#qi-hardware IRC log for Monday, 2010-10-04

wpwrak_the ghost is mysterious indeed. the cache has some subtle changes in the postscript. when i run things manually, there are no changes. interesting.00:01
wolfspraulcan you reproduce it on your end already?00:02
wolfspraulI noticed the first ghosts appear with Adam's second commit already, so I will cut out everything else and only run those 2 revisions.00:03
wpwrak_i have one ghost but i'm not sure it's the same as yours00:03
wpwrak_you have of course a whole castle full of ghosts :)00:04
wpwrak_in my case, there are single-pixel differences in the ppm0. in the corresponding postscript, there are some subtle changes like line thickness. however, when i generate the postscript manually, there are no differences.00:05
wpwrak_i'm not regenerating the whole thing but i keep all the checked out revisions. then i can run things directly there and see what happens00:06
wpwrak_generated. now let's see what happened ...00:15
wpwrakoops, killing dbus-launch has interesting side-effects00:18
wpwrak(ghostbusting) aha ! now i get different ps. one step closer00:20
wpwrakoh dear00:26
wolfspraulyou are ahead of me. so far I only found that when I go back to say the 2nd commit, no ghosts. but they will show up (in that same commit) if I include later commits.00:26
wpwrakin one version, the R comes from the kicad "device" library, in the other from the cache00:27
wpwrakso it'a basically a difference of local kicad installations that shows up there.00:28
wpwrakcarlos' local version leaked via the cache.lib that he also checked into git00:28
wolfspraulmaybe sanitize the cache.lib as well?00:30
wpwrakso the difference is real all the way. it just has a weird origin.00:30
wpwrakhmm, delete it ? ;-)00:30
wpwraklet's see what happens ...00:35
wolfspraulI don't think files leaked from Carlos can explain it entirely00:50
wolfspraulmaybe Adam also leaked files00:50
wolfspraul(if it's that)00:50
wolfspraulcheck this: http://projects.qi-hardware.com/schhist/board-qi-avt2/00:50
wolfspraulwhen I git reset upwards, and then run schhist, this is the first time I see ghosts00:50
wolfspraulat that point there are no commits from Carlos at all yet00:51
wpwrakwell, files leaked by whoever committed the cache.lib. kicad takes care of making sure the leakage occurs :)00:51
wolfspraulin my url, if I run schhist up to and including the third commit (from the bottom), it's still fine00:52
wolfspraulbut when I let it run up to the 4th (visible) commit, the ghost diffs in the 3rd (visible) commit appear00:52
wpwrakah right, there's also stuff from adam. i just wrote "carlos" because i saw him there. doesn't really matter who committed, though.00:52
wolfspraulok should I wait until you commit further sanitizing, or continue digging deeper on my own?00:52
wpwraklet's first see what the cache removal brings when it comes to the one ghost i had. rebuild is running.00:53
wpwraki suspect that there may be more amiss in your case, because the change keeps coming back, but maybe not00:55
wpwrakoh ... could it be that the change keeps repeating when committers alternate ? one version with adam's cache, the next with carlos', etc.00:55
wpwrakno. that doesn't make sense.00:56
wpwrakcaches should get synchronized as well. hmm.00:56
wolfspraulI focus on the first time any problem appears, and then I would dig in there.00:57
wolfspraulbut it's on hold until your cache sanitizer improvements00:57
wpwrakhmm. it gets worse. if i delete the cache, it doesn't find the components anymore.00:58
wpwraklayouts will be fun. just spotted some nice long local absolute paths in the profile :-(00:59
wpwrakokay, the missing components are caused by the origin having been removed from the profile01:01
wpwrakso the project only works if you have the cache01:01
wpwrak(or add them back)01:02
wpwrakwe can do this, of course ...01:03
wolfspraulyou lost me know01:05
wolfspraulwhich origins are removed?01:05
wolfspraulI'd say if something is missing, let's just not show a diff then, rather than work with unstable cache files that are most likely only accidentally committed01:05
wolfsprauls/lost me know/lost me now/01:06
wpwrakthe profile (.pro) lists the libraries we use. local ones and system ones. in this case, the profile was cleaned up at some point in time.01:06
wpwrakwhen that happened, the reference to the "device" library was removed. that one has things like R. so with the cache.lib now gone too, kicad no longer knows where to find this symbol01:07
wolfspraulbut why do the ghosts show up on the server but not your machine?01:10
wolfspraulwe just need to run exactly your setup on the server and we're done :-)01:11
wpwrakmaybe different local libraries. at least before the cache was checked in, kicad would use the local libs.01:11
wpwraki still don't know why your ghost is a revenant, but the stuff is so messy that there may just be a path i'm not thinking of at the moment01:12
wolfspraulI took the Debian 20100314 KiCad package, and added your patches to it01:13
wpwrakbut lemme add the libs. the you can try and see how much this improves01:14
wolfspraulso I did not build directly from KiCad upstream, but from the Debian package as a starting point01:14
wpwraknext build is running ...01:22
wpwrakhmm. better, but now they disappear in one version and come back a bit later. odd.01:40
wpwrakah, it's different parts that do the vanishing game01:43
wpwrakokay, these are just profile editing mistakes. they were so bad that they were eventually noticed and reverted despite the cache covering things up01:48
wpwrakso, hopefully final test01:58
qi-bot[commit] Werner Almesberger: Added mmone-jtag-serial-cable. http://qi-hw.com/p/eda-tools/8be9fe102:08
qi-bot[commit] Werner Almesberger: gitsch2ps now by default removes cache.lib (option -c keeps it) http://qi-hw.com/p/eda-tools/7b1d3fd02:08
qi-bot[commit] Werner Almesberger: Environment variable SCHHIST_ADDLIBS to force libraries to be present. http://qi-hw.com/p/eda-tools/4d28daf02:08
wpwrakalright. export SCHHIST_ADDLIBS=device02:08
wpwrakand it will make sure we always have at least this02:08
wpwraklet see how it goes02:09
wolfsprauldo I need to add SCHHIST_ADDLIBS into my script as well?02:14
wpwrakat least for avt2, yes02:15
wpwrakhaven't tried the rest yet02:15
qwebirc3299hi xiangfu02:18
xiangfuqwebirc3299: Hi02:18
qwebirc3299I have forwarded the schematic of level converter to Adam02:18
xiangfuqwebirc3299: yes. I think he can help you.02:19
qwebirc3299is there any possiblility of prob around kernel ported to NN02:19
qwebirc3299i have downloaded the latest  image from the website02:20
xiangfuqwebirc3299: http://downloads.qi-hardware.com/software/images/NanoNote/Ben/testing/2010-09-14/openwrt-xburst-qi_lb60-uImage.bin,02:23
xiangfuqwebirc3299: this one is more recently. include the 'setfont2' patch. which make the terminal font more smooth.02:24
xiangfuqwebirc3299: for now. the kernel pretty stable.02:24
qwebirc3299so shall I port this and then try out02:25
qwebirc3299is it something different vis-a-vis serial prot02:26
xiangfuqwebirc3299: if you more focus in serial console. no need to try out the 2010-09-14 kernel.02:28
xiangfus/focus in/focus on/02:28
wpwrakwolfspraul: btw, for deanonymizing qwebircs, one possibility would be to put the nick in the real name field if nickserv rejects it02:30
wpwrakwolfspraul: this is what my schhist of avt2 now looks like: http://downloads.qi-hardware.com/people/werner/schhist/avt2/02:32
qwebirc3299xiangfu , is Adam trying to get some soln for my prob02:58
xiangfuqwebirc3299: hmm.. you want create your own converter. or you just want the output from Nanonote?03:02
qwebirc3299anything which can give me the reqd output03:02
xiangfuqwebirc3299: then we can try to send email to mailing list. see if someone near you have bought a work converter :)03:04
xiangfuqwebirc3299: or you can easy to buy from amozon ...03:05
qwebirc3299the schematic i have sent to you and Adam, will it work or not ?03:05
xiangfuqwebirc3299: sorry. I don't know much about the hardware schematic. I don't understand the low level hardware stuff.03:07
Action: xiangfu only know how to read write GPIO :)03:07
qwebirc3299ok that is software part03:08
SiENcEhi bartbes. nlove works good so far. i made a small game with it. still some problems using some functions, but all in all it works.03:12
kyaklol, nice /etc/banner03:57
kyakwe should also have a custom image during boot03:58
SiENcEbartbes ?07:52
wolfspraulwpwrak: strange, I still get the same ghost diffs as before, as if SCHHIST_ADDLIB didn't do anything07:59
methril_workwow!! imoressive the schematics history08:40
wolfspraulmethril_work: yes, but we are still fighting some bugs08:46
wolfspraulbut yes, it's great!!!08:46
wolfspraulmethril_work: you can find them all here http://en.qi-hardware.com/schhist08:46
wolfspraulah sorry, wrong link :-)08:47
methril_worki'll love08:47
wolfspraulI hope this will also create some discipline among our KiCad committers towards clean commits, clean commitlogs, etc.08:47
wolfspraulnow that their work is dragged out in the open so much more :-)08:48
methril_workand it's only highlighting the modified part!!08:48
methril_worki never see that on schematics08:49
methril_workwolfspraul, i ended paying my work hours (more or less)08:50
methril_workso i'm going to have some time now!!! :)08:50
wolfspraulyou mean you quit your job?08:51
methril_workno, i quit my extra hours (for the moment)08:52
methril_workso i only have to work 9h at day08:52
wolfspraulah OK, well great! welcome to our project!08:52
wolfspraulman we do need you :-)08:52
methril_worki'm going to have 2/3 spare hours at time08:52
wolfspraulwhat are you most interested in?08:52
methril_worki'm interested in almost everything, hehehe08:53
methril_workbut i'm going to work on RTAI 4 MIPS processors (for the moment) if you didn't have any priority task in mind08:54
methril_worki love the Milkymist08:54
wolfspraulRTAI 4 MIPS? what is that?08:59
thevishyis it possible t replace a laptop monitor with a different model ?08:59
thevishyi dont mind if it doesnt fit properly to my laptop case08:59
thevishylike say mine is 14w , would 15w fit ?09:00
methril_workReal Time09:01
wolfspraulmethril_work: how does it relate to other things like Milkymist, SIE, etc?09:03
wolfspraulor the other way round - how can we help with your project?09:03
wolfspraulwhat's the goal?09:03
methril_workwolfspraul, it's related with SIE, to get a real time api forrobotics09:04
methril_works/forrobotics/for robotics/09:04
methril_workand also could use scilab with rtai extensions to get a simulink/labview similar environment with SIE (mostly)09:05
rafamethril_work: wow.. you have a 9h work per day and also you can work 2 or 3 hours more. Cool ! :) .. I can just work around 6 hours per day for all (paid+hobby), no more, maybe less. maybe i have some kind of lazy disease :P09:06
wolfspraulmethril_work: do you have a SIE board?09:07
methril_workrafa, well, it depends how you organize your time ;)09:07
methril_workwolfspraul, not jet09:07
wolfspraulwhere will you get it from?09:07
rafamethril_work: yes, something of that could be the reason09:07
methril_worki'm waiting for a Carlos sent :)09:07
wolfsprauli thought the plan was for several months to get one to you...09:07
methril_workwolfspraul, it was09:08
methril_workbut it didn't arrive09:08
methril_workbut i could start this port with the ben09:08
wolfspraulyes good09:08
methril_workdo you think there are most important tasks to deal with?09:08
wolfspraulthanks for asking09:09
wolfspraulI think the main thing is that you find something that truly interests you, that helps you in the future09:09
wolfsprauland then, second most important, that you are steadily working on it, not just a burst for a few weeks then nothing for months or longer09:10
methril_workwell, i'm also interested in milkymist, but i've to study a lot to get a productive level09:10
wolfspraulyes sure09:10
methril_workso i see it like a long term plan09:10
wolfspraulSIE is a good start09:10
wolfsprauland NanoNote is a start for SIE09:10
wolfspraulthat's why we have all these projects together here :-)09:10
wpwrakwolfspraul: (ghosts) i was afraid yours wouldn't go away so easily :-(09:11
wpwrakwolfspraul: btw, SCHHIST_ADDLIBS only corrects a problem unconvered by removing the cache. removing the cache is what should stabilize the process. the cache is now removed by default.09:11
methril_workrafa, you live in Argentina, isn't it?09:12
wolfspraulrafa: speaking about SAKC/SIE boards, what happened to that one professor you gave one to?09:13
rafawolfspraul: hey, it was sakc.. he started to port jlime to sakc and work on the bootloader for sd, then also there was a student interested to work his thesis with something about next SIe or Sakc. But no idea currently. I need to ask why he is quiet nowaday..09:16
rafawolfspraul: well, I started my trip and forgot the world for a month or two as well :D09:16
rafamethril_work: yes, where are you from?09:17
methril_worki'm from spain but i'm living in Brazil09:17
methril_workin the south of brazil (Rio Grand do Sul)09:17
rafamethril_work: ah.. we are neighbors09:17
methril_workrafa, yes!! :)09:18
rafamethril_work: you are almost in uruguay or argentina :)09:18
methril_workrafa, i'm, I would like to plan a trip to Uruguay, or maybe Argentina09:18
methril_workrafa, i'll ask you for advices ;)09:18
rafamethril_work: if you come to argentina please keep in touch09:20
rafamethril_work: we can meet here09:20
methril_workrafa, i'll do :) Where are you? in Buenos Aires?09:21
rafamethril_work: of course, no doubts.. please tell us if you are coming some day.. we know buenos aires a lot.. so you will have fun if you ask us .. Also we can talk about copyleft hardware and things to do together perhaps09:21
wpwraksoon, we'll have enough developers in south america that we can benefit from the inferiority complexes of the political class and get government funding ;-) a state of the art pan-continental technology project where something actually gets done. that's definitely something new around here :)09:21
methril_workwpwrak, this is new around the world ;)09:22
wpwrakmethril_work: in europe, you get some EU-sponsored R&D projects that sometimes do get something done (ACTS and the like). of course, that whole thing is rather expensive for its output. not that i'd complain too much - it served us well back then ;-)09:25
wpwrakwolfspraul: do i have access to the machine where the schhist process lives ? may be easier if i can just log in and grab the files.09:27
methril_workwpwrak, i know, i was living in Europe until the beginning of this year09:27
wolfspraulwpwrak: try ssh root@09:27
wolfspraulroot = access, no? :-)09:27
wolfspraulhostname is fidelio.qi-hardware.com09:27
wpwraki'm beginning to worry about the security of my own systems :)09:27
wpwrakprojects. ... works. too ;) thanks !09:28
wpwrakah no,. wrong server. darn, no locked doors anywhere :)09:29
wolfspraulyes, it's not the same server09:33
wolfspraulthe buildhost (fidelio) is basically the building sandbox09:33
wolfspraulalso it's not backed up09:33
wolfspraulsystem is tuned for performance (raid0 etc)09:34
wolfspraul100% of the data on it is recreatable09:34
wolfspraulso for schhist, fidelio is ideal09:34
wolfspraulonce I do my wiki->openwrt packaging system, the build steps will also run there. everything else would be crazy insecure.09:35
wpwrakthe happy sandbox farm :)09:35
bartbesSiENcE: you expect me to reply while I'm gone?09:36
wpwrakdownloading the critters ...09:36
bartbesanyway, what problems?09:36
wpwrakbartbes: awol ? tsk tsk. we have to send igor over to tighten these chains again.09:36
bartbeswpwrak: I was KIA09:38
wpwrakbartbes: ah one of these immortals again. we didn't have that in our files yet.09:40
bartbeswell, obviously you went for the wrong intelligence agency09:41
wpwrakrafa: btw, i briefly discussed with wolfspraul the jlime vs. qi-hardware situation, namely that jlime doesn't get much exposure/visibility due to the patent issues.09:41
wpwrakrafa: i also remember that you weren't quite sure what to do about it. correct ?09:42
wpwrakbartbes: i'll have those traitors executed. can't have that if your goal is world domination. now, where's my white cat.09:43
bartbeswpwrak: I think I saw it killing some dogs over there09:44
kristianpaulwolfspraul: wiki -> openwrt ,  how is that?09:48
SiENcEbartbes, hey. well now you are here. i just waited09:49
kristianpaulsomething like make a package as you write a wiki article? :)09:49
rafawpwrak: I am sure that we (jlime) should do to fix the patent issues (on software side), but that I am not sure is how to work with other jlime devs it. We have a lead/dictator/main developer, kristoffer, but he wants, like me, decide what to do deciding between all devs (we are really few guys). But then, we discussed it at jlime meeting..09:49
SiENcEbartbes, just want to say that nlove works on dingoo a320. on weekend some friends and i wrote an love2d game and i ported it to nlove (just because not all functions work in nlove)09:50
SiENcEbartbes, i can send it to you later09:50
rafawpwrak: .. OE already includes all this kind of stuff (software/libs which can encode/decode patented stuff)..09:50
wpwrakrafa: so some didn't like the idea of making it easy to exclude patent-encumbered packages ?09:51
rafawpwrak: and we should manage that there. Currently, another jlime dev is working with OE for muffinman/nn, and we could work that. Buuut,09:51
wpwrakwolfspraul: actually, at what level does the PATENT flag work ? package yes/no, build process changes, patches of packages ?09:51
SiENcEbartbes : i just want to ask, if you plan to merge nlove with love2d ... and if you plan to finish it?  it would be good to have an git or svn repository for the code09:52
rafawhen we were in jlime meeting, we realize that tuxbrain decided to use jlime anyway.09:52
rafawpwrak: and wolfspraul already said that tuxbrain was the most concerned part of patented issues.09:52
wpwraktuxbrain_away: so you're no longer worried about patent trolls ? or just sleeping badly every since that fateful decision ? :)09:53
rafawpwrak: so we decided to wait a while before to do the hard work.09:53
rafawpwrak: I guess that tuxbrain is really worried, but he did not have a lot of stuff to decide what to use09:53
rafawpwrak: so we will wait a while to know what other nn resellers like to have inside of nn09:54
wpwrakrafa: fair enough. of course, the patent problem will hit everyone in the distribution process. tuxbrain is just the most exposed at the moment, being in europe, the home of the mp3 trolls.09:54
rafawpwrak: we are the dark side so far :)09:54
rafawpwrak: you will not find jlime info if you are a qi user who just check qi wiki .. but if you look deeper .. ;)09:56
wpwrakyes, the lack of information is what worries me a bit. of the thre choices, openwrt, jlime, and debian, jlime seems to be the one best adapted to the device. yet it's kinda hard to find.09:57
wpwrakso many people will just assume that the default (openwrt) is implicitly recommended, stay with it, and maybe start fixing it and porting stuff to it, things you've already done long ago on jlime09:58
wpwrakthus, instead of making a good distribution even better, all this work goes into making a weaker distribution stronger. so the average does improve, but the maximum doesn't.09:59
kristianpaulopenwrt will stay there for devel, but jlime is looking endusers now >)09:59
wpwrakkristianpaul: i'm currently fairly distribution-agnostic (my user space is the shell :) so i haven't tried jlime yet. is it in any way less convenient for developers ? e.g., is it harder to get a host toolchain ?10:00
kristianpaulwpwrak: (jlime devel) no no,my point was in eduser experience development not development tools support10:00
kristianpaulbut thats other point for jlime10:01
wpwrakwolfspraul: btw, do you know what license xue is under ?10:01
bartbesSiENcE: I'll probably merge it in after 0.7.0 releases10:02
wolfspraulwpwrak: gpl/cc-by-sa10:02
wolfspraulthe files are not as clean as I wish, we need to clarify in more places10:03
wolfspraulwpwrak: just look here http://en.qi-hardware.com/wiki/Xue10:03
wpwrakwolfspraul: ah, perfect. thanks !10:03
rafawpwrak: I guess that one nice thing with our stuff is the complete repositofy built from OE. If you use jlime you do not need to spend a lot of days to have some application ported, you just install it from repository. The same with debian. But also, we have worked with tiny screen, 16MB of ram, slow cpu devices for several years, so we alredy suffered testing a lot of different kind of stuff for those devices. That is why we know that still if gtk1 is rea10:04
wpwrakwolfspraul: regarding the PATENT flag, at what granularity does it work. e.g., for mplayer, does it just kick out mplayer, does it pick a different build script / change options in the existing script, does it apply special patches to the source, ... ?10:05
wolfspraulkristianpaul: yeah I just want something that lets you hook into the buildhost from the wiki, say write a script in the wiki, have it packaged on the backend10:05
wolfspraulI'll do the minimum plumbing necessary to hook the wiki & buildhost/openwrt build environment together10:05
rafawpwrak: also.. we already know how to config the light X which use just few 5 MB of ram or less.. and which wm (we alredy tested around 20 these years)10:05
rafato use10:05
wpwrakrafa: OE is a mixed blessing :) it gives you a lot of stuff but at the same time it's quite a monster10:06
wpwrakrafa: so it's quite hard for new people to work in there. openwrt has an advantage in this regard.10:06
rafawpwrak: yes, it is a monster, but if you use it just to build your repository it willl be quiet complete.. and you will have all the stuff already built for your arch.10:06
rafawpwrak: ah.. you means to expand it?.. it is already okey :)10:07
rafano add more stuff please!10:07
kristianpaulindeed about monster10:07
kristianpauli wonder how works bitbake inside10:07
wpwrakrafa: reminds me of that guy who said around 1900 that all interesting things have been researched and he didn't expect significant any further technological discoveries/inventions :)10:08
wpwrakkristianpaul: probably much like the digestive tract of the title character of "alien"10:09
wpwrakhmm, that message needs reshuffling. i shouldn't multitask10:09
wolfspraulwpwrak: patented flag - don't know exactly. the flag is passed into the Makefile, I guess it could be passed deeper if necessary.10:10
wolfspraulremember that openwrt is a patch management system10:10
wolfspraulso the patches can of course drive this flag deeper into upstream sources10:10
aisaGood morning hackers.10:10
wpwrakrafa: a high barrier for participation means more reliance on the people who have already mastered that barrier. now, in the case of jlime, that's unfortunately only a small group. so there is a continuity concern.10:10
aisaOr good evening for those of you who experienced Monday already.  :-D10:11
wpwrakwolfspraul: okay. so what happens with you set that flag, in terms of differences in the output ?10:11
SiENcEbartbes, good. is there a forum thread about nlove? or what about an coderepository? i had to patch some functions to get it working10:12
wolfspraulwpwrak: remove mp3, mpeg4, h.26410:12
wolfspraulaisa: sorry I'm slow. the news stuff we talked about is not ready yet. give me another day.10:12
SiENcEbartbes : i still have the problem, that starting nlove alone couses an segfault. but starting nlove with pong for example works10:12
wolfspraulfeel free to start editing of course, no hesitations...10:12
wpwrakrafa: i remember OE well from openmoko. we had that problem that lots of people had to wrestle with it, with all the problems this causes. things got a bit better when there were more or less dedicated maintainers, so one could just drop off a package and they would take care of the OE-specific side.10:13
tuxbrain_awayregarding jlime and pantent trolls and tuxbrain10:13
wpwrakrafa: those working on GUI-ish things still worked a lot in OE, though. not sure how well that went.10:13
tuxbrainI manually remove mp3 codecs on the Nanowar edition of jlime10:14
wpwrakwolfspraul: do you remember that ? people working on OM GUI apps using OE as their build environment ?10:14
aisawolfspraul: it is fine with me, Monday is difficult for me: too many things to get done.10:14
aisaindeed!  I will.10:14
tuxbrainso at least at by default it will not play them10:14
wolfspraulwpwrak: I remember clearly. That's why I (that is just me, personally), chose to try OpenWrt this time around... :-) Nothing against OE though, and especially not a nice distribution like Jlime.10:14
wpwrakwolfspraul: (mp3, ...) okay, so find granularity. good.10:15
wolfspraulbut please, nobody ever try to force me back into OE/bitbake again10:15
aisaIf Jlime dealt with the patent issue, is that the only reason stopping adoption of it?10:15
wolfspraulneither directly, nor indirectly (managing people who have the suicidal mission to go there)10:15
rafawpwrak: well, yes and no. we are a small group.. but OE is being worked for many groups. I am not sure that you means with a monster, I like OE to build a complete repository, I do not like so much it to keep different root filesystems. For repository it is nice.. you just has a tiny.bb for an application and it builds, like you had in gentoo for example. And OE is quiete complete.. so you do not need to add many applications to the feed, and others also d10:15
wolfspraulnothing stops 'adoption'10:15
wolfspraulalready now10:15
tuxbrainalsa: for my part yes10:15
wolfspraulI reflashed the new run of 1000 nanos recently10:15
kristianpaulwolfspraul: i think rafa already  did (sacrify) for us :)10:16
wolfsprauland we didn't have a good openwrt build (the one we ended up flashing has serious bugs)10:16
aisaI guess I'm troubled by this: I'd like to dive deeply into a distribution and make it something really special on the NanoNote.10:16
wpwrakwolfspraul: i know what you mean. i've seen the OE wars for something like a year even before you joined :) (don't even remember how long ... certainly felt like half a lifetime)10:16
wolfspraulI was pretty close to just reflashing jlime without asking anyone :-)10:16
wolfspraulbut then I thought it's too chaotic...10:16
aisaWith no other information, I wish to use the one that comes with my NanoNote, in the hope that my work also makes it into the official image.10:16
tuxbrainif jlime was pantent savy I will ask wolfgang to send my NN with Jlime instead of openwrt10:16
tuxbrainin spite it sound cruel10:16
tuxbrainafter the 1000 reflashing10:16
wolfspraulno problem, I'd love to do that10:16
aisaBut it seems like much energy is applied toward Jlime or other environments,10:17
wolfspraulif my customers, someone who wants to buy, challenge me, as a manufacture, to put software A/B/C on it, that's a great challenge10:17
aisaand that OpenWrt, while being the image that comes flashed on the device, receives the least attention.10:17
wolfspraulthat's the beauty of free software10:17
wolfspraulI will gladly accept such challenges, and get the stuff reflashed in the way people want it. done.10:17
aisatuxbrain: who is Mr/Mrs Jlime?  I want to talk to them.10:17
wolfspraulaisa: OpenWrt receives a lot of attention10:18
kristianpaulaisa: you meant then muffin man?10:18
tuxbrainalsa, the user experience with jlime worht that energy10:18
rafawolfspraul: if debian would have some kind of repositories for mobile devices (with small, just minimal files packages for every application) we would use debian surely..10:18
wolfspraulyou will not be able to marry the two here, many people tried before you :-)10:18
wpwrakaisa: funny, to me it seems that actually quite a lot of work goes into openwrt :)10:18
wolfspraulthere are good reasons for both of them10:18
tuxbrainat least on my side as one who has to buy it10:18
rafaaisa: you can join us in #jlime channel10:18
bartbesSiENcE: that is because I guess the a320 can't handle fb windows larger than its screen size10:18
tuxbrainI love the simplicity of developing on Openwrt10:18
rafaaisa: you can talk there with us .. and you would meet MR JLIME! :D10:18
bartbesSiENcE: and no, there's no repo atm, but fixes should be submitted to me10:19
aisamy knees shake!  :-D10:19
wpwrakrafa: has any attempt been made to change debian in this direction ? at least to the point of getting the opinion of the "core" debian maintainers ?10:19
tuxbrainI have succesfully make my own builds and adaptations, thing that I hade never fully achieved with OE on openmoko era10:19
wolfspraulaisa: OpenWrt & OpenEmbedded are very different10:19
kristianpaulrafa: is that too bad emdebian?10:20
wolfspraulthe judgment will of course differ depending on which side of the camp you stand on10:20
aisaI was around when OE was formed/created.10:20
wolfspraulit's clear that OpenEmbedded is as much real free software as OpenWrt is10:20
aisabut I haven't looked at it since then.10:20
wolfspraulputting some border cases like binary firmware blobs and patents aside, we all feel similarly on those anyway.10:21
wolfspraulfor me personally, I'm now sold on OpenWrt10:21
SiENcEbartbes, :-( ok. i can send you a patch. but it would be good to know where i can find updates for nlove src. if sdlsound is working10:21
wolfspraulbut I'd love to reflash any software, preferably free software, on NanoNotes to sell them10:21
kristianpauli like openwrt simplicy as far i can see10:21
wpwrakrafa: the things is that OE is very very complex. for openwrt, if you understand makefile and kconfig, you're almost there. for oe, you need to learn about bitbake, all its configuration files, and sometimes also things like caches and such. (the latter mainly because OE moves a lot of stuff. that's of course a scalability issue also openwrt will run into, if brought to the same number of packages)10:22
wolfspraulthe next big thing will be the software for Milkymist One and then Xue, the digital camera10:22
rafawpwrak: no.. we do not have any contact with them (debian), no sure about emdebian as kristianpaul says.. We should check how emdebian is going.. I am not sure if it is very used.10:22
wolfspraulif you would let me choose, which tool is best for that task, I would pick OpenWrt over OE many times10:23
aisawolfspraul: what is the one feature of OE that OpenWrt doesn't have?  Since you would pick OpenWrt, what would you miss about OE?10:23
bartbesSiENcE: I guess I could blog about new releases :P10:23
wolfspraulOpenEmbedded has a much larger repository though, and much more history & packages on the GUI side10:23
tuxbrainwolfspraul: I agree10:23
wolfspraulI wouldn't miss anything from OE.10:23
wpwrakwolfspraul: from a system developer point of view, i liked the "lots of packages (provided by someone else)" plus my root-making script plus an easily installed host toolchain that can also install libraries (with opkg)10:24
kristianpaulwolfspraul: wpwrak in openmoko what was the experience with debian in that time?10:24
SiENcEbartbes : ok :-) . i would also port it to caanoo10:24
bartbesSiENcE: though I might put it up on bitbucket next time I work on it10:24
SiENcEgph caanoo10:24
wpwrakwolfspraul: the workflow is then like on a regular linux pc - you rarely build things your actual work is based on yourself, you just install the package10:24
wolfspraulanyway we are not talking about the tool OpenWrt vs. the tool OpenEmbedded now, or rather that's a separate discussion10:25
wolfspraulwe are talking about Jlime10:25
SiENcEbartbes : bitbucket is ok . better would be github but ok10:25
wpwrakkristianpaul: (debian) i think a bit like here - it exists but never got quite that much attention10:25
wolfspraulJlime is a great distribution for NanoNote, definitely10:25
bartbesSiENcE: well, love is on bitbucket10:25
SiENcEah ok10:25
bartbesalso, with the new owner bitbucket got cooler10:25
SiENcEi have to test it10:26
bartbesand promises to get even cooler10:26
aisaWill OpenWrt scale to the number of packages to make the NN a nice working environment?10:26
tuxbrainaisa: matter of time and matainers but yes it should10:26
aisaDoes that mean I can go clean up the Utilities submenu in make menuconfig...10:27
aisait is screens and screens long.10:27
kristianpauljlime is great as rafa point dont get overloaded with too much features10:27
SiENcEbartbes : i think boot.lua has to be patched to work on nanonote and dingoo. lower standard resolution an better no vsync10:27
wpwrakwolfspraul: i think there are several issues to consider with jlime or any other distro: one it the end user experience, one is how well it fits the developer's need, then whether it's long-term viable, which breaks down into a) current maintainers, b) accessibility for new maintainers, and c) alignment with other work (i.e., how much actually needs to be done to keep it afloat)10:27
kristianpaulaisa: i think the idea of openwrt simplicy is just have what you need10:27
aisawpwrak: +1.  Very well said.10:28
SiENcEbartbes, but all in all ...its really cool to code in love2d and than use the same code to run it on handhelds :)10:28
aisaI worry that Jlime won't go anywhere unless the NN is aligned with it, and doing so may mean both get to a dead-end.10:28
wpwrakwolfspraul: the jlime vs. debian issue is also relevant. rafa mentioned to me before that he'd actually prefer jlime to be debian-based, but the issue is that debian doesn't have the right granularity for small devices10:29
aisaWe are very open to new developers, so if we branch, I think that problem is the same regardless of distribution.10:29
aisasave issues with upstream.10:29
wpwrakaisa: (dead end) that's also my concern10:29
SiENcEbartbes, last weekend we where on a hobby game development meeting and we took part on an overnight game coding contest. since the last 4 years we took Love2d for this...and last year we won the 1st price.10:29
aisaI believe that xiangfu and wolfspraul were trying to tell me some of that last night, but it is much clearer to me this morning, after reading what you said.10:29
kristianpaulwpwrak: talking about GUI end user experience openmoko moved to Qt isnt?10:30
kristianpaulas they found gtk(2?) was too slow10:30
kristianpaulbut now jlime shows somthing diferente10:30
wpwrakaisa: ah ... no, i misread what you wrote :) no, i'm not too worried about both going into dead-end land so far. i'm worried about jlime not going anywhere without closer ties to the nanonote.10:30
aisawpwrak: take the dead-end issue to its logical conclusion, should we not align with the largest embedded handheld linux distribution?10:30
aisaas a space, it isn't going anywhere.10:30
tuxbrainuser experience(jlime wins) developer's need(mmmm draw, OE more libraries, openwrt more easy to manage), lonterm- viable...... if we don't sell more NN  who cares :)10:30
aisawpwrak: ah, ok.  I agree.10:31
aisatuxbrain: well said.  What sells more NN?10:31
aisadoes our choice of distribution matter for that?10:31
wpwrakaisa: and the nanonote doesn't look its best only with openwrt. so there's a nice symbiosis there. nn gives jlime a reason to exist and nn lives a happier life if jlime is around.10:31
SiENcEbartbes, btw. i'm a close friend of qubodup10:31
wolfspraulwpwrak: yes!10:31
SiENcEreal life10:32
wolfspraulaisa: there is no 'choice of distribution', also no 'alignment'10:32
aisawolfspraul: I am trying to hear you, but I'm troubled by the size of our community and what I see as all of us going in separate directions.10:32
bartbesSiENcE: yeah, but since 0.7.0 features a new no-game screen anyway.. I was lazy10:32
aisaso when I read that, what I hear is that we'll be as flexible and accomodating to everyone,10:32
bartbesSiENcE: that is cool10:32
aisaand I fear in that won't create a great experience for anyone.10:33
bartbesSiENcE: and that is even cooler10:33
aisaI want to be convinced out of this, stop worrying, and start working.10:33
aisabut I'm not there yet... :-(10:33
wpwraktuxbrain: what i'm saying is that i, as a system developer, was happy with the OE situation we had at the end in OM. the badness of OE was nicely shielded from me and the made pretty packages for anything i wanted.10:33
bartbesSiENcE: btw, my slow replies are because I'm watching a vid fullscreen, so I don't get notifs10:33
wpwraktuxbrain: so i didn't have to touch OE myself. hence the issue of OE complexity vanishes. that is, as long as there's someone who maintains it for me. i do occasionally like outsourcing ;-)10:34
SiENcEbartbes, np :) .  So what i want to say is, that i'm going to support nlove or love2d for handdevices where i can!10:35
wolfspraulaisa: yes I understand10:35
wpwrakaisa: (too many directions) yup. that's what troubles me with openwrt. there are probably around ten people doing distro maintenance work at the moment, and they're split into three directions already.10:35
SiENcEbartbes: and i really want to port our games to this devices10:35
tuxbrainaisa: don't get wrong or worried, qi-comunity is on of the most pushing in one direction one I have ever meet :)10:36
wolfspraulaisa: whenever I learn about some new free software, or let's say often, I think "oh no, why can't we do this with software XYZ"10:36
aisatuxbrain: well, I came to the right place then.10:36
wolfspraulbut then, when you look deeper, you see the reasons10:36
tuxbrainmost of us are happily using at time debian/jlime/openwrt :)10:36
wpwrakaisa: this wouldn't matter if they're only focussed on a single distro anyway, i.e., they would refuse to touch anything else. but i think most of them would be happy to contribute anywhere, so it would be beneficial to steer them into the direction where their work has the largest value.10:36
wolfspraulOpenEmbedded will not go away, it's far too big and diverse10:36
SiENcEbartbes: lalalove, "28 brainz later", Zurückgefilmt, ...10:36
wolfspraulvery very diverse actually10:36
bartbesSiENcE: I guess when 0.7.0 releases I will fork it on bitbucket and start merging in the nlove changes10:36
bartbesSiENcE: I have only seen lalalove really10:37
wolfspraulneither will OpenWrt, which differs in many ways from OE10:37
wolfspraulneither will Debian10:37
wolfspraulneither will Fedora10:37
aisathat means Jlime is the only open question.10:37
kristianpauliris? :)10:37
aisaon this issue.10:37
aisakristianpaul: hehe.10:37
SiENcEbartbes, oh you know it :) ... well we are going to make a website with all 4 games10:37
wolfsprauland I also expect more interesting microkernels to appear in coming years, actually we have a nice little one on the NanoNote already, Iris (not very functional though)10:37
wolfspraulaisa: open question?10:38
aisawolfspraul: question of long-term viability.10:38
wolfspraulJlime is today probably giving you the best end user experience on the Ben NanoNote.10:38
wolfspraulgreat long term viability10:38
wolfspraulI know rafa, I'm not worried.10:38
wolfspraulhe's a strong man!10:38
wolfsprauland there are others too10:38
wolfspraulwhich question are we trying to answer now, actually?10:39
tuxbrainso rafa: will jlime community will do somthing about the pattent for I can struggle wolfgang to sendme jlime NN or not ?10:39
wolfspraulI lost track a bit.10:39
aisaIf I had to summarize this conversation right now, I'd say we use OpenWrt because wolfspraul likes it and is familiar with it.10:39
tuxbrainbtw rafa you own me a ogv player :)10:39
aisaAnd that we have lots of options/directions as a community for platforms, for better or for worse.10:39
wolfspraulhe :-)10:39
wolfspraulmy superpowers10:39
wpwrakwolfspraul: i'm a bit worried about him deciding to get a job that saps too much of his strength :)10:39
wolfspraulaisa: it's not what I like10:39
wolfspraulOpenWrt has done a great service for us10:40
wolfspraullet's just take one thing out - Linux kernel upstream work10:40
aisaThat is much better said, thank you.10:40
wolfspraulthe NanoNote was released in February this year10:40
aisaI'm sorry it took me so long to get here...10:40
wolfspraulin a few weeks, when 2.6.36 comes out, we will have 90% of the Ben NanoNote kernel in kernel.org10:40
wolfspraulthat's awesome!10:40
aisavery, very awesome.10:40
wolfspraultake that for 'long term viability' :-)10:41
wpwrakhow are openwrt and kernel upstream connected ? kernel should be pretty distro-agnostic10:41
wolfspraulwhich tool helped us to get there?10:41
aisaThis is a much clearer picture for me.10:41
wpwrakyou mean "having any userspace at all? ?10:41
aisait also ties into our conversation last night, wolfspraul, where you articulated other reasons things are the way they are right now.10:41
wolfspraulOpenWrt is really strong on kernel patch management, and the people that have helped us drive the kernel upstream are essentially using openwrt as their development/patch management system10:41
wolfspraulthe question is not what could theoretically happen10:42
kristianpaulwolfspraul: uboot too10:42
wolfspraulof course theoretically this could also happen with OE, or buildroot, or whatever10:42
wolfspraulbut it did not, and it did happen via OpenWrt10:42
aisabut in actual fact, it is happening right now with OpenWrt.10:42
wpwrakrafa: what's your feeling about jlime and oe vs. debian. do you think a switch is something that's likely to be attempted ? do you know what the other jlime developers think about this ?10:42
wolfspraulaisa: correct10:42
wolfspraulwpwrak: I think Debian on the NanoNote is so slow that it is unusable, unless you are very very generous10:43
aisaok, pretend the opposite for a moment.  What does Jlime have or do that OpenWrt doesn't?10:43
wpwrakwolfspraul: making it not so slow would be part of the challenge :)10:43
aisaWhy all the Jlime love?10:43
Action: wpwrak gives the word to tuxbrain :)10:44
aisaLike, actual Jlime love, not "here is what I hope."10:44
kristianpaulbecause run X nicelly and rafa support us too much :)10:44
aisawhy do you today install Jlime instead of OpenWrt?10:44
aisakristianpaul: And OpenWrt?  Not as nice?10:44
aisaor just is it the work hasn't been done yet?10:45
tuxbraindue it works out of the box smoth and easy10:45
aisaI'm asking: social or technical barrier to solving this?10:45
tuxbrainwith a good bunch of apps by defoult10:45
kristianpaulbut  i must said last openwrt image looks better10:45
aisawonderful wonderful.  This is much clearer to me now.10:45
kristianpaulgood :)10:45
aisaYou're all very awesome.10:46
tuxbrainI allways said the ideal NN OS is Openwrt boottimes + Jlime user experience and package aviable10:46
aisaMan, boottime matters insanely much to me.10:46
tuxbrainthen go for Openwrt10:46
aisawell yeah, but then I'm going to add all this crap.10:47
aisasorry, end-user experience.10:47
aisaand it will go get all slow.10:47
wpwrakaisa: well, you've perhaps seen tuxbrain's nanowar promo video. the device with jlime basically looks like something commercially made for regular users. the ben with openwrt greets the tinkerer, promising lots of problems that need fixing.10:47
wolfspraulopenwrt has been designed with target platforms with 4 mb or 8 mb or ram originally10:47
wolfspraulit may even run on 2mb targets10:47
wolfspraulthat's basically the bottom end of the 'openwrt range' (it's optimized for)10:48
wpwrakaisa: nothing wrong with the latter per se, since tinkerers are what we want, but it also changes the focus of people10:48
wolfspraulthe NanoNote with 32mb has a nice decent amount of memory for openwrt10:48
wolfspraulI think OE is good for 64-256 mb memory10:48
wpwrakaisa: instead of doing new things with the platform, they work on catching up with the minimum status quo10:48
wolfsprauldebian or fedora above that10:48
wolfspraulbut guess what - Fedora is coming down10:48
wolfspraulif the NanoNote had 64 MB, Fedora would have started supporting it already (so I am told)10:48
kristianpaulaisa: also jlime key bindings are more generic than in openwrt10:48
wolfspraulFedora has now declared 64 mb to be their 'bottom end'10:49
wpwrakaisa: so instead of building spacecraft they dig up the ground to put some sewage system10:49
tuxbrainI just want costumers damnit, not thikeres or such , for me they can usit to equilibrate a table meanwhili it buy them for me </shopkeeper mode off>10:49
wolfspraulI think OpenWrt is the stronger tool if you want to make a custom image10:50
wolfspraulsay a customer wants to order 500 NanoNote, but put a total customized image on it10:50
tuxbrainwolfspraul: I agree again10:50
wolfspraulwould you point them to OE or OpenWrt?10:50
wolfspraulI would point to OpenWrt many times over10:50
wpwrakwolfspraul: i would use something like myroot, fed by a rich package repository10:50
wpwrakwolfspraul: why even bother with the build process if someone else has done it already for you anyway ?10:51
wolfspraulwpwrak: fine, but you get my point. if the choice is between owrt or oe, in that case you would also recomment owrt.10:51
wolfspraulJlime has great images today, and a great package repository.10:52
wolfspraulsome patent issues though10:52
wpwrakwolfspraul: as i said, i was happy with the arrangement we had at the end in OM. both from an "it worked for me" point of view as well as from a conceptual point of view.10:52
rafawpwrak: no, we have not done anything about debian, we like it because it builds, as you know, repositories for different kind of archs.. and those are really huge. But we do not have idea if they will have something nice for small devices in the future. If they do we could use that someday (and to help them of course).10:53
wpwrakwolfspraul: so i'm looking for ways to get closer to something like that. the questions whether openwrt or OE is easier to build only matters if you're working on that sewage system. if you're building the spacecraft, you'd prefer to not touch either at that level.10:53
wolfspraulrafa: I think Debian will never work really well on 32 mb10:53
wolfspraulservers, desktop, notebooks all go up10:54
wolfspraul8 gb is normal nowadays10:54
wolfspraulon servers > 20 gb10:54
wpwrakwolfspraul: to answer your question, if i had to make that heavily customized image myself, then my choice would be either openwrt or port myroot and run it off the jlime packages (if that's possible).10:54
wolfspraulservers will have more than 100gb memory soon10:54
wolfsprauleven for our devices, we don't have 32 mb because we like artificial restrictions10:55
aisaI'm logged into a 16-proc, 132gb linux box right now.10:55
wolfspraulit's because it was the best / most high-tech device we believed we could tackle with our resources10:55
wpwrakwolfspraul: if i need that heavily customized image but i'd get external help, it would depend mainly on who's best at solving my needs. and there jlime looks like the better bet. (depends of course on what exactly needs customizing)10:55
wolfsprauland it seems we can, barely... after 1 year :-)10:55
wolfspraulI wish we could go to 64 mb, 128 mb, etc. the more the better.10:55
wolfspraulno 10 horses will drag me back to bitbake10:56
rafawpwrak: for the rootfs image I like just the minimal rootfs that OE builds, but I do not like how OE keeps different kind of distributions/images. So far, I just build the minimal rootfs with OE and repository, then just a makefile or script adds some extra packages and multimedia stuff. A jlime dev is making the whole image inside of OE like it should be if you use it. But well, for the rootfs image I just like a simple makefile/script to build that :)10:56
wolfspraulI only have so much time in my life. enough for that one...10:56
Action: aisa amused10:56
wpwrakrafa: could the first step be to more clearly identify the technical reasons for it being difficult for small devices ?10:56
Action: kristianpaul curios about the wpwrak myroot syste,10:57
wolfspraulaisa: yeah, really. I suffered for 1.5 yr from it. I had a whole team on this beast, 3 or 4 full-time people.10:57
rafawolfspraul: yeah.. debian is for PC only I guess10:57
wolfspraulif wpwrak says he would pick that for the custom image, he he. I will watch gladly, sipping my drinks...10:57
wolfspraulno thanks10:57
aisawolfspraul: this would suggest that we could in theory catch up and surpass OE, by being more nimble.10:58
wolfspraulJlime images - great! getting into the bitbake mess myself again - NO.10:58
aisa*unless* we give up that nimbleness because it isn't possible at the scale OE operates at.10:58
wpwrakkristianpaul: http://svn.openmoko.org/developers/werner/myroot/10:58
wolfspraulhard to say, and that is definitely no my/our problem10:58
aisathis much is good to hear.10:58
wolfspraulI'm focusing on all things hardware, manufacturing. so much to free up there.10:58
wolfspraulboth OE and OpenWrt are quite active projects10:59
wpwrakwolfspraul: yup, packages is what i'm after. and nice people like zecke who add stuff i ask them for within hours :)10:59
wolfspraulyou cannot pick a 'winner', no matter how hard you try, because they are very different once you spend some time diving in.10:59
rafawpwrak: we should write that at wiki maybe.. so the technical reasons are modified there. debian packages are big, for example those brings man pages, extra docs, etc. Also there are packages with multiple files for different kind of stuff that a small device could not need. For example perl with around 1000 files.. a small device could have just few disk space to install the whole perl and adds. Maybe you just want to use perl and some extra perl adds to 11:00
wolfspraulaisa: openwrt has about 2000 packages now, openembedded maybe 15,000 or so11:01
wolfspraulnot that these numbers mean much11:01
wolfspraulI wish there were 100 apps that work REALLY WELL on the NanoNote, in either owrt or oe/jlime11:01
wpwrakaisa: i'm not sure openwrt scales to that size and functionality. you very often hear people complain that X is way too complex and slow etc. so they build Y that only implements 10% of what X can do and is really nice clean and fast. then they encounter more and more of the real life demands X has already faced in the past and they add things to Y. eventually, Y is just as sluggish and ugly as X and people start complaining about Y be11:02
wpwraking so horrible, make their own leaner and cleaner Z, etc.11:02
rafawpwrak: then, you also needs the rootfs image.. maybe with busybox, dropbear, etc.. or something light.. I am not sure if debian already builds those minimal rootfs images for different kind of archs11:02
wolfspraulowrt will in general be able to build customized rootfses easier, be able to target devices with less ram than oe11:02
wolfspraulI think owrt is the better environment to prepare patches for upstreaming, although someone who uses oe to do that may disagree11:03
wpwrakrafa: (package granularity) is there a reason why packages could not be broken down into smaller pieces ? too much work ? too complex to do ? opposition from the "core" debian people ?11:03
wolfsprauljlime are really cool end-user ready images for Ben NanoNote, and guess what, they use OE! :-)11:03
wolfspraulso you go figure what all this means11:03
wolfspraulDebian runs as well, but slow IMO11:03
wolfspraulFedora is interested, but 32 MB is below their minimum requirement of 64 MB, and the few 64 MB boards we have were not enough to get them started11:04
wpwrakrafa: i wouldn't use the distro per se to make the rootfs images. thigs like myroot can do this already and they're nicely layered on top of the distro.11:04
wpwrakrafa: so the distro makes the packages and nothing else. then you have a rootfs maker than picks the things you want for your device, including UI style and what not.11:05
wolfspraulwpwrak: yes but your logic is not always automatically true11:05
wolfspraulsometimes an effort to do everything 'leaner' than the old stuff will only realize later why all that old stuff was needed11:05
wolfspraulbut sometimes not11:05
wpwrakwolfspraul: yes, sometimes someone succeeds in actually being better than all those before :)11:05
wolfspraulsee the big success of lighttpd, for example11:05
wolfspraul(just random example, I hope this doesn't create another long Apache pro/con discussion now...)11:06
wolfspraulI think OE is a bit older than openwrt11:06
wolfspraulmost owrt people at the begining were ex-OE (I'm vague on this, might be wrong)11:06
wolfspraulI hope OE, and especially Jlime, is a happy family member of the NanoNote family11:07
wpwrakwolfspraul: hmm, i use awhttpd. i suppose that doesn't put me in the camp of enthusiastic apache supporters ;-)11:07
rafawpwrak: I would do the same like you.. I would like to see a complete repository of packages for small devices, with the last versions of open source software, like debian has for PC. Then I would use something like your rootfs maker which picks stuff from the repository to build a complete image.11:07
wolfspraulif not, we need to do more to make them happy11:07
wpwrakwolfspraul: alright, one step at a time. let's first make them happy. so for you to be happy with making them happy, they have to make tuxbrain happy by avoiding the patented stuff.11:08
wolfspraulbefore my next reflashing round, I want to apply the lessons learnt in the last 1k reflash action11:09
wolfspraulfor that I think I need a few weeks at least11:09
wolfspraulneed to look into some initramfs issues, and some usb hug related bugs11:10
aisaThis reminds me, I have just updated my NN to the latest tracking_backfire,11:10
aisaand I expected to find many more bugs than I did find.11:10
aisaCan someone remind me of the problem we're having in the current image?11:10
wpwrakwolfspraul: (btw, my reason for pushing towards a more feature-complete disto is that i consider the ben basically as a pc. in fact, except for the screen resolution, it runs circles around my first linux pc. and that was a "serious" workstation at that time.)11:10
aisaI thought it was things like 'strace --help' crashing.11:10
wolfspraulSDL not working11:10
wolfspraulif you boot, try starting dgClock, for example11:11
aisafrom the command-line?  got it.11:11
wolfspraulNanoMap is qt-based, I think it works11:11
wolfspraulcommand line?11:11
aisaRather than gmenu2x.11:11
wolfspraulyou built your own image? well I'm not exactly sure then what you selected.11:11
aisaI'm running config.full_system on my NN right now.11:11
wolfspraulremember this: NanoMap is qt-based, imgv uses SDL11:12
aisaisn't qt SDL backed?11:12
wpwrakrafa: so, to remove these codecs from jlime, would you know what to do ? would the other jlime people agree with having such an option ? would this option be something that could be merged into OE upstream ? and, last but not least, is there anyone who'd actually have the time and skills to do it ? :)11:12
wolfspraulif you can get both to work, the bug is fixed11:12
aisaor is our qt running X?11:12
aisagot it.  NanoMap & imgv must both work, or there is a bug :-)11:12
wolfspraulconfig.full_system uses a framebuffer, no X11:12
wolfspraulso qt also on fb11:12
wpwrakrafa, wolfspraul: one issue with codec removal: for those who want the questionable codecs, what would be an acceptable way to provide them ?11:12
wolfspraulthat's a legal question11:13
wpwrakwolfspraul: that's why i'm asking you ;-)11:13
wolfspraulto the best of my knowledge, _links_ is what will set them off11:13
wolfspraulso let's say a distributor in Europe is selling the device11:13
wolfspraulthe device has a URL 'qi-hardware.com' printed on it11:13
wolfspraulnothing on the qi-hardware.com or affiliated servers must point towards the patented stuff11:14
wolfspraulor the other way roudn - the people running the qi-hardware.com servers have the responsibility to avoid patent infringement11:14
wpwrakincluding irc or mailing list archives ?11:14
wolfspraulfor a forum, this can mean to remove posts that explain how to do it11:14
wolfspraulfor a wiki, same11:14
wolfspraulyes sure11:14
wolfspraulit's a legal thing11:14
wolfspraulit must be reasonably doable of course11:15
wpwrakso it's active or passive (by not acting) endorsement11:15
wolfspraulI wouldn't push it though. if the wiki homepage has a big link "how to enable mp3", and you think you can get away with "we would have removed it after notification from sisvel, it's a wiki", that's very very risky11:15
wolfspraulthat's all legal stuff11:15
wolfspraulyou cannot be 100% sure until after the whole court case is over11:16
wpwrakyeah, that would be too obvious :)11:16
wolfspraulanother option would be to distance the device from qi11:16
wolfspraul'unbrand it'11:16
wolfspraulthe thinking always starts with the last guy selling (actually importing) the device into a patent-infected country11:16
wpwrakbut that wouldn't help tuxbrain who sells a bundle, would it ?11:17
wolfspraulthat guy has a legal responsibility to not infringe, and to not cowork with people who infringe11:17
wolfspraulimagine the device is totally no-name11:17
wolfspraultuxbrain could say they have nothing to do with qi-hardware.com11:17
wpwrakyup. like the pulster incident11:17
wolfspraulof course then there must be no link from tuxbrain.com to qi-hardware.com, anywhere11:17
wolfspraulbut right now, with qi-hardware.com printed on the device, these ideas are worthless anyway11:18
wolfsprauland tuxbrain.com has tons of links to Qi11:18
wolfsprauland if Tuxbrain would really have to remove all this, and stop talking about it, I'm also not so sure how this could work out11:18
wpwrakyup. this may just create too much confusion11:18
wolfspraulat some point it really does become underground, and then the MP3 'capability' also looses its value11:18
wpwrakwhen will the patents expire ? in ~2 years ?11:19
qi-bot[commit] Xiangfu Liu: add 'rm -f fped' to make clean http://qi-hw.com/p/fped/9e2ee4511:19
qi-bot[commit] Xiangfu Liu: add debian package stuff http://qi-hw.com/p/fped/731491f11:19
wolfspraulaccording to wikipedia in a few years, yes11:19
wolfspraulI doubt sisvel will do much to enlighten anybody about this11:19
wolfspraulin fact they will probably increase their scare tactics towards the end11:19
wpwrakbtw, regarding patent expiration. you mentioned that MIPS vs. china case. do you know what went wrong there ?11:19
wolfspraulanyway, that's speculation11:19
wolfspraulI don't know when they expire.11:20
mthbut isn't it the device manufacturer that can be forced to license patents rather than the hardware designer?11:20
wpwraki suppose it will be headline news when they do. then wait a bit to see if there are any traps.11:20
mthif I understood your explanation a while ago, Qi Hardware doesn't manufacture11:20
wolfspraulyou think a Chinese company can be forced to do anything? especially from a Western company?11:20
wolfspraul80% of the world are not patent infected11:20
wpwrakmth: i think it's basically that you can attack whoever is in the chain.11:21
wolfspraulso this is a system some rich countries created because they think it helps them11:21
rafawpwrak: maybe one easy fix could be removing the codecs for main rootfs image, and the system should not have any link to download packages or visit official web sites. If the user wants codecs he will need to check the web. If that would be useful then we would not need to merge that into OE. We just build the image (with OE perhaps) and then we fix it with some "clean" task/script/whatever. If you want to know if we could do a work like : one OE version 11:21
mththey won't pressure the Chinese company, but they'll try to stop the devices from being imported into a country which does enforce patents11:21
wolfspraulmth: yes right, but we are talking about legal responsibilities now.11:21
wolfspraultuxbrain sells11:21
wolfsprauldevice has 'qi-hardware.com' printed on it11:22
wolfsprauluser enters 'qi-hardware.com', and is greeted by big announcement "click here to learn how to install mp3"11:22
wolfspraulthat would be a pretty easy case for sisvel11:22
wpwraka system some lawyers in rich countries maintain because their villas, sports cars, and alimony for ex-trophy-wives prove that it helps them. somehow :)11:22
wolfspraultuxbrain would be the one to either license the patents, or refuse to resell that product until the manufacturer does11:23
wolfspraulmth: yes correct.11:23
wpwrakrafa: so that would be basically what tuxbrain did, just automated11:23
mthwolfspraul: I mean someone could use Qi's design and sell a device with MP3 capabilities by installing a modified firmware or giving the users an easy way to do so11:24
mthputting precompiled binaries on qi-hardware.com is probably asking for trouble11:24
mtheven if you'd win in the end, it would be time and money consuming11:24
wolfsprauleven sources, although the most important are instructions11:25
rafawpwrak: maybe we could provide two rootfs image versions: current one and another one with "clean" stuff for resellers :)11:25
wolfspraulthink from a normal end user, they try to plug that hole11:25
mthafaik the LAME project didn't have problems spreading sources11:25
wolfspraulI need to go sleep11:26
wolfspraulthe patent industry has learnt11:26
wolfspraulno more gif11:26
wolfspraulwhy go after broke college kids which will only make you unpopular11:26
wolfspraulso of course all free software is 'exempt' nowadays11:26
wolfspraulha ha11:26
wolfspraulno money there anyway!11:27
wpwrakso ther ehas to be a clear organizational separation between qi-hw and tuxbrain on one side and jlime on the other. that is, as long as jlime hosts patented packages or sources.11:27
rafamth: the problem is not the software I guess, the seller selling machines with that kind of software will get into troubles11:27
wolfsprauland even better, the free software kids now help them spread their technology to people who have money, and who are worth milking11:27
wolfsprauland in the end the college kids pay again, because the tax is built into any electronic they buy11:27
wolfspraulmuch much better than the gif chaos11:27
wolfspraulwpwrak: correct that was the old plan11:28
mthrafa: yes, someone spreading device + software will have to pay for the patents in some countries, until we get more sane laws or all relevant patents have expired11:28
wpwrakor keep jlime perfectly clean and have an underground jlime-pirates organization that makes those packages11:28
mthI don't know about the USA, but in Europe pure software is not patented, but some combinations of device + software are11:28
wpwrakmaybe assange's looking for a job soon :)11:29
wolfspraula patent is a patent, there is no need to put a big title "software patent" on it11:29
wpwrakwolfspraul: would that old plan still work ?11:29
rafawpwrak: do you know what do other companies do? for example IIRC dell was selling machines with ubuntu.. They just pay money to do that?11:29
wolfspraulit needs to be connected to a machine, so it can only be enforced once some piece of hardware comes into play11:30
wolfspraulcanonical is paying11:30
rafawpwrak: what would you like to see?.. for example.. if you sell machines.. you will not be able to sell machines in USA or europe if you want to sell those with debian (for example)11:30
wolfspraulof course only for them, i.e. if you put the same ubuntu on a machine and sell it they will come after you still...11:30
wpwrakrafa: unless you or your upstream or your distributors pays the patent parasites11:31
rafawpwrak: and how could we use debian?.. would that be useful?.. debian is huge.. and it would be a mess to try to use it without codecs11:31
wpwrakrafa: why would debian minus codecs be more of a mess than jlime minus codecs ?11:32
wolfspraulwpwrak: paying them is not that easy actually11:32
wolfspraulthat's another story11:32
wolfspraulthey will demand that you take the license for all your products, for their future innovations, and so on and so on11:32
wolfspraulwhatever they can squeeze into you, and they have thought about that for a long time11:32
rafawpwrak: well, debian is bigger I guess and a lot more of people involved than oe11:32
wpwrakwolfspraul: lovely :) where are they ? palermo ? :)11:33
wolfspraulplus a 6-figure down-payment into their account, in case you go bankrupt11:33
wolfsprauland so on11:33
wpwrakso what happened in the pulster incident ? did you ever get these devices out of customs ?11:33
wolfspraulI guess they get away with all this legally.11:33
qi-bot[commit] Xiangfu Liu: add ITP bug number to 'debian/changelog' http://qi-hw.com/p/fped/b0062ee11:34
wolfspraulfor some magic reason this is not called 'extortion'11:34
wpwrakrafa: ah, you mean a bigger mess to get people to agree on the plan. well, perhaps. i wonder if this isn't on their radar already. debian tends to be very sensitive about legal issues.11:34
wolfspraulwpwrak: I think in the end yes. I'm sure pulster had to sign all sorts of crazy agreements, need to ask him if he wants to talk about it...11:34
wpwrakwolfspraul: just like patenting something isn't called plain old theft, even though it is11:35
wolfspraulwpwrak: yes but Debian is software only, 'exempt'11:36
wolfspraul(not very generous considering that software-only infringements cannot be enforced, and the infringers would not have enough money to be worth going after anyway)11:36
wpwrakwolfspraul: yes, but it hits anyone bundling debian with hardware. so it would be reasonable for them to care11:36
wolfspraulthe 'free software is exempt' strategy works quite well11:37
wolfspraulyou can try to argue with them :-)11:37
wolfspraulI was very happy that OpenWrt did have a soft ear for our troubles.11:37
wpwrakyou tried  ? so they don't understand the issue ? or they do but still don't care ?11:37
wolfspraulYou can read between the lines that Jlime/OE is struggling with the same issues.11:37
wolfspraul"is this really our problem?"11:37
wolfspraul"software patents don't exist"11:37
wolfspraul"free software is exempt, look here what they say..."11:37
wolfspraulwell I tried successfully with Owrt, done11:38
wolfspraulwith OE (Jlime) - in process - see above11:38
wolfspraulDebian, no, never tried11:38
wpwraki'm not sure this is the issue with jlime. i think the situation is pretty clear. but people may simply not be aware of it.11:38
wolfspraulyes but it's hard to get people to join11:38
wolfspraulalso because patents are such a mess11:38
wpwrakso, where's a debian lobbyist when we need one ? :)11:38
wolfspraulnew enemies may pop up out of thin air11:38
wolfspraulit's depressing to think about it logically11:39
wolfspraulit is not logical11:39
wolfspraulsame as a street robbery is not logical11:39
wolfspraulI just need mp3/mpeg4/h.264 removed right now11:39
wolfspraulmore 'as needed' :-)11:39
wolfspraulif you are very rich, there are many many more11:39
wolfspraulsee the nonsense the large corps have to pay all the time, like RIM, Microsoft, etc.11:39
wolfspraulthe only thing that protects Tuxbrain from most patents is that it has no money11:40
wpwrakit's a pity that they closed that loophole that artificial islands can no longer become souvereign states. otherwise, i'd have a nice revenge phantasy.11:42
wolfspraulpatents only affect about 20% of people11:42
wolfspraulit's a social system for lawyers11:42
wolfspraulthere are other social systems as well, so let's just get on with it :-)11:43
wpwrakcreate two long thin islands in the atlantic and the pacific. build tall wall, too high to fly over them. implement laws that treat involvement with patent much like some asian and middle eastern coutries trear illegal drugs. build nice airports to tunnel through the wall, us-style with immigration and emigration. put the names from the various patent databases in that countries' terrorist database. wait for the first planes to land an11:44
wpwrakd have the firing squads ready.11:44
wolfspraulwpwrak: now we outlasted anybody else with this conversation! -)11:44
wolfspraulI need to go sleep...11:44
wpwraklet's hope something good crawls out of it :)11:45
wpwrak(the conversation)11:45
wolfspraulI think we have a fairly solid plan.11:45
wolfsprauljust need to see how/where Jlime fits.11:45
wolfspraulthere are multiple spots, let's see what works best for them11:45
wolfspraulthen we'll do that11:45
wpwrakyup. it's all about sustainability11:46
wolfspraulaisa: hope you are not shocked about this endless conversation11:51
wolfspraulwpwrak and I get carried away sometimes11:51
wolfspraulI'll try to finish the news stuff in the wiki tomorrow, then we can go from there11:51
wpwrakrafa: so ... has all this helped, to clarify things or otherwise ? :)11:58
aisawolfspraul: I was in a meeting.12:14
aisaI'll have to catch up to know whether I'm shocked.12:14
rafawpwrak: yes, it clarified why qi can not have any mention about jlime I guess. Now, I would do the next proposal to jlime: we write a script which picks the jlime image and does a new jlime rootfs image clean of codecs, clean of jlime links. Maybe that is useful for resellers, and also some users could know that jlime has a distribution for nn.12:14
rafawpwrak: no sure if we should remove the stuff from our repositories and OE. I mean, as example, for users like me, I like to use Debian like it is .. so I guess that we like to use jlime like it is. But on the other hand if qi is the only start point for all the people with nn,12:16
rafathen it can be hard for users to know if there are options to use.12:16
wpwrakrafa: i'm actually not sure if qi-hw would have to avoid mentioning jlime. it may be sufficient not to mention the problem packages.12:17
wpwrakrafa: it all really depends on what wolfgang and tuxbrain fell comfortable with. as wolfgang has said, there's no guarantee that you won't be attacked no matter how careful you are12:19
rafawpwrak: that I really would like to avoid is, and that is a concern for me is: if some new user wants to use nautilus, and he only wants to test it.. then today he will try to port allllllll the libs nautilus needs to openwrt. If he wants to help with openwrt then that is great and I love all that hard work, because I would do that. BUt, if he just wanted to test that.....12:23
rafaman.. I have seen some people trying to build stuff for many days.. just because they wanted to test.. and they just were thinking that it is the only way they have.12:23
rafaThat is not nice I guess.12:23
wpwrakyes. i think quite a lot of that sort of repetitions is happening here. people get the ben, see that openwrt is the distro, find their favourite package is missing, and then work on porting it.12:25
rafait is just an example. Still if nautilus will not be useful on nn.. why he would suffer many days trying to build stuff just to test. I would like that he knows that some distributions already have packages to install12:25
wpwrakwhile that makes openwrt stronger, it may draw development resources from more interesting tasks12:25
rafayes, I would like that users know both things : that if they just want to test/use stuff there are options to use packages already built. And that if they want to help to openwrt then they need to work to do openwrt stronget, which is a great thing .12:27
aisaSince our Asian developers presumably just went to bed.  (Or should have) who is left?12:30
aisaIt is still morning for me, and I'd like a sense of who hangs out here during my daytime,12:31
aisaso I know who to pester.12:31
aisaI'm working on packages and the base system for the NanoNote, and I'm in the Southwestern USA.12:31
wpwrakbuenos aires, 13:30 here. brunch time soonish :)12:35
aisawpwrak: You're not the only one down that way, I believe.12:45
qi-bot[commit] carlos: Adding serial flash to FPGA http://qi-hw.com/p/nn-usb-fpga/d9b9e5212:45
wpwrakaisa: naw, not by far. e.g., rafa lives just a few blocks away12:45
aisaI used to work with two developers in Montevideo.12:46
aisawpwrak: What kinds of things do you work on here?12:48
aisarafa: same question to you, what are you working on here?12:49
wpwrakaisa: right now, qi-hw is my fulltime hobby :)12:49
aisaVery nice.  What portion do you specialize in?12:50
aisaor do you do it all?  :-)12:50
wpwrakaisa: before, kernel and general low-level stuff at openmoko. before that, other kernel stuff. never for local clients, though. not much technical work going on here, i'm afraid12:50
rafaaisa: :) ... right now qi-sw is almost my fulltime hobby.. jlime is my full time :)12:50
wpwrakluckily, we have the internet :)12:50
aisarafa: Are you hoping to see Jlime become an "official NanoNote image" or are you happy with how things are now?12:51
rafaaisa: I need a related job.. so if you know just tell me !12:51
aisaI used to work at a company that hired globally.  I'm not currently, not right now anyway.  I will remember this anyway.12:52
rafaaisa: official nanonote image is not our business, so we are happy so far. But it would be more useful if we have more users as well.12:53
rafaaisa: but we are okey and enjoy how it is. We prefer to be choosed because people like it12:53
aisagood perspective, I had not thought of that.12:53
wpwrakrafa: there's no such thing as "free choice" :) whatever ships with the device will have a big headstart13:15
aisaIndeed, most users will not modify the defaults.13:16
aisaI think this holds true ever for advanced users.13:16
aisaThough clearly at some point it doesn't, because some people run Jlime.13:16
wpwrakrafa: also pushing the choice to the customer isn't entirely fair. people will still need some guidance. very very few will possess the information necessary to decide that on their own at that time.13:16
wpwrak(and those probably don't care much anyway because they would also know how to switch to a different distro)13:17
rafawpwrak: yes, but I was guessing that (that we prefer people choosing jlime because they like it) because if you use linux surely like to try other distributions/flavors, and also the project started saying that nanonote current users are developers. But I have seeing that no.. that users just use the default distribution and they are not thinking, in a big % of them, if there is or not another option13:22
rafawe will see ..13:23
kristofferwpwrak, aisa we cannot influence what software qi want to put on their machines, so we have to fight on quality rather than having a head start.13:27
wpwrakrafa: yes, but that's pretty much what is to be expected :)13:29
wpwrakkristoffer: but we can influence it :) and why not combine quality and the head start ? :)13:29
wpwrakaisa: some people will always experiment. but many just have some specific use in mind. or just don't have the energy/time to change distros (even if it's easy)13:30
kristofferwpwrak, that would be perfect, but as I said its not up to us (<-jlime)13:31
aisakristoffer: I suspect if you handed wolfspraul an image, he'd flash devices for you, up to what you expected to be able to sell.13:31
aisaIs that really the limiter?13:31
kristofferaisa, most likely true, but that would also imply that the buyers would be aware of any other option than openwrt. Which isnt the case.13:32
wpwrakaisa: imagine. you get the device, experiment around a bit. maybe overcome some obstacles. eventually, you'll realize that the default choice is not the best. but then you'll also have solved the most pressing problems for you. and you'll have invested time into customizing the system. so when you're at the point of making the decision, it will be harder to implement it.13:32
aisaThis is true.13:32
aisaI feel like I'm there right now.13:32
aisaI spent the weekend learning enough about my NanoNote to feel comfortable with the device,13:32
wpwrakaisa: (limiter) what ?13:33
aisaand that I can deeply explore.13:33
aisaAnd I'm concerned that I'll invest a lot of effort into OpenWrt and find I'd rather have done that for Jlime.13:33
wpwrakaisa: and at the same time, you may be afraid that jlime may be a dead end because its not officially supported, and thus your effort there would be wasted. very tricky all that13:34
aisawpwrak: Limiter being: Jlime images aren't and won't be available through sharism.cc13:34
aisawpwrak: Yes.13:34
wpwrakof course, i understand why wolfgang fears OE. OE was our basis at openmoko. it was unspeakably painful.13:35
aisaThis was good to hear this morning.13:36
aisaAs I did not know why we weren't on OE.13:36
wpwrak(limiter) well, as long as the patent issues aren't resolved on the jlime side, wolfgang may not feel comfortable to have much in terms of pointers on qi-hw at all13:36
wpwrakso jlime will actually be hard to find13:36
wpwraki think one of the problems with the use of OE in OM was that developers were expected to run it. what was missing was a group of maintainers acting as a buffer13:37
wpwraki mean, on my pc, i run ubuntu. but i wouldn't even know (without looking) how to build one of their packages from sources13:38
wpwrakso if the complexity of OE is hidden, all is fine. naturally, hiding the complexity has a cost of its own13:39
wpwrake.g., you need to have people who do that "now", and you have to make sure there will be people who do that also in the future13:39
aisaI can't quite tell the difference between Jlime and OpenWrt in this regard.13:41
wpwrakin OM, people on GUI applications actually used bitbake to build their own software, just as if it was "make". that was because they didn't want to have to worry about library dependencies (of which there were many for those packages) and they were probably also afraid of the (largely imagined) complexity of doing a cross-compilation13:41
aisaIs Jlime it's own platform?  Based on OE?  How does it work to build an image?13:41
wpwrakthen of course, when things went wrong, it wasn't just a small typo in a makefile but some weird failure somewhere in a bitbake run. etc. chaos and mayhem.13:42
wpwrak(maintenance) openwrt is more accessible, so the risk is lower13:42
wpwrakwith jlime/OE, if a meteorite hits the next jlime meeting or if ~2 people lose interest, the whole thing comes apart. so it's very fragile.13:43
wpwrakand of course, by keeping it marginalized, that doesn't change. chicken and egg again.13:44
aisaI was just thinking that.13:44
aisathat fact scares me, which makes me not want to participate, which contributes to the problem.13:44
wpwrakif jlime was the "mainstream", one or two people may actually master the complexity of OE and could act as maintainers. that would greatly lower the maintenance risk13:45
wpwraki don't know who does how much with the ben in jlime. rafa is very active but i don't know about the others.13:46
aisaFrom this conversation this morning, I see that I need to learn more about jlime, but I'm more certain I'll stay with OpenWrt.13:50
aisawhich is a funny position to be in,13:50
aisawanting to get more information but feeling more certain about my decision :/13:50
wpwrakcertainty is a good thing - while it lasts :)13:51
kristofferhey, Im right here :) please stay to facts if you are going to rant on us/me13:52
aisafact check!13:53
wpwrakkristoffer: when i wrote "the others", i was in fact mainly thinking of you ;-) are you working on the ben, too ?13:53
kristofferwpwrak, not currently since Im stuck with bootloader work on j7xx atm.13:54
wpwrakkristoffer: but in general yes then. good :)13:54
kristoffer1. first off, simply removing the copyrighted codecs wont magically make jlime into an option13:54
kristoffer2. openwrt is nice for small systems that need to be 100% reliable. OE is nice for machines you are planning to actually use13:55
wpwrak(1) why not ?13:55
aisaI have that question too.13:55
kristofferwpwrak, yet again im not the person to ask13:55
wpwrak(2) agree that OE is much more featureful and more suitable for this kind of device. as i said, i view it as a PC, not as an embedded dish washer controller or such13:56
wpwrak(regarding 1) and the issue is mainly patents. not sure if there are copyright issues as well.13:58
kyakopenwrt is limited in many ways, but it is still enough for me13:58
kristofferI wont go into any OE vs openwrt match really, it all comes down to that we are working from the outside and will need to put out a good product for others to use it. So that is what we are trying todo.13:59
wpwrakkyak: you're actually one of the people who make me feel that precious resources are being squandered. just curious, did you come to openwrt with the ben or did you work on openwrt before ? (now my question would be embarrassing if you are some founder/celebrity of openwrt ;-)14:01
rafaaisa: one of our goals is to have everything ready to use at jlime.. for example.. you said that you have used several days learning enought about current software in your nn to feel comfortable. Why? is it hard?.. For us, we want to install -> use14:02
wpwrakkristoffer: i think OE vs. openwrt is a secondary issue. right now it's openwrt more or less out of the box (with qi-hw working closely with openwrt upstream) vs. jlime14:02
aisarafa: It is because I want to actively develop, so going from source code to compiled and tested code on my NN required me to learn.14:03
wpwrakkristoffer: so if jlime was based on openwrt, with a similar number of packages and a similarly polished interface, why not. in fact, it's good if this isn't visible for anyone but the distribution maintainers.14:03
rafaaisa: I see.. and what do you want to develop?14:03
wpwrakkristoffer: the only issue there is what allows them to get their work done and to make sure there will be distribution maintainers to hide all this from the rest14:04
aisarafa: version 1.0 is to port some Lojban-related software.  My longer term vision is to have a workflow for writing and performing material.14:05
aisaI'm working on a longer description of that, that is brief...14:06
kyakwpwrak: you are flattering me.. i first used openwrt on my Linksys WRT54GL router (and still using), maybe that's why i found it exciting to play with it on Ben14:06
kristofferwpwrak, I think it all comes down to us prefering OE as a development platform, and with our own GUI (which I like to think is more intuitive). I think you describe it well when you say that end users should just be able to do their work and not think much about how the underlying software actually works. We will also keep closely synced with OE to make sure all work gets merged.14:07
rafawpwrak: we are few guys there nowm and I hope that any bomb/meteorite hits next jlime meeting :) .. but we are there since 2006 maybe.. and now at least four of us have nns.. I am not sure if there are a lot of people in qi working doing openwrt better. Also, remember that we use OE, and if I read well OE has a lot more people working for different kind of mobile devices than openwrt..14:08
rafaaisa: wpwrak: so I am not sure if openwrt for nn will be very active in the future.. Are the commits for openwrt coming here as qi-bot?.. if yes.. i do not see a lot of work being made. You can tell me.. well, if qi does not develop openwrt for nn anymore it is becaues nn failed and maybe no more devices will be built.. but then, better for jlime.. I am sure that we will continue improving the software for the current devices sold14:09
kyaki think it's great that one can choose between distributions14:10
kyaki myself sometimes boot in jlime to check things there14:10
rafakyak: yeah.. and I would like to see more.. but of course.. no many nns around yet14:10
kyakrafa: https://dev.openwrt.org/browser actual openwrt development14:11
kyakwe are just syncing with them from time to time14:11
kyak(if i'm right)14:11
wpwrakkristoffer: (end users) also most developers i think. just consider how often you become aware of the underlying build process on your desktop linux system. almost nobody strays beyond the package manager.14:12
rafathat any= that no any :P14:13
wpwrakrafa: not sure how much we see here of openwrt development. i'm sure there's more ben-relevant stuff also in upstream but i don't know if anything specifically made for the ben shows up first here or if people also commit directly into upstream.14:13
wpwrakrafa: also, there's pretty much zero jlime development activity visible on this channel ;-)14:14
rafahaha.. well, actually I just see your commits.. it looks like you are the only one working these days :)14:14
kyakif it's not his job, i'd say he's the only one NOT working these days :)14:15
kristofferwpwrak, we try not to spam qi-hard.. with our development. We try and limit it to announcements and the occasional problem that arises.14:16
kyakwpwrak: really, how do you have so much time to devote to ben development?14:16
wpwrakkristoffer: yup, that good. i was just making fun of rafa's metric :)14:17
kristofferwpwrak, :D14:17
wpwrakkyak: oh, continuing a bit with the habit of pretending i can afford it :) i spent about a year trying to get a successor for the openmoko freerunner off the ground. so what's a few more months.14:18
wpwrakeventually, material reality will catch up with me too, though. i'm already getting more careful about not putting too many things on my plate.14:19
kyaka hungry engineer is a terrible thing :)14:21
kyakunless your plate is metaphorical14:21
wpwrakfood is cheap and plentyful in argentina :) so it's the metaphorical plate :)14:23
rafaaisa: interesting.. (about Lojban).. well, when you want to start please tell us.. we have different ways to have software built for jlime.. so if that is useful for more people we can help you to have it into jlime as well (different ways: we build packages using OE, but users can build packages using a jlime toolchain or if they are traveling they can build software in nn as well.. installing gcc, make, and friends..)14:48
aisarafa: thank you.  I have one package build, but not fully useful yet.  And a couple more packages I'll be porting in the short time from now.14:50
kyakhm, i wonder why openwrt's gcc for compiling on target depends on TARGET_x8615:05
kyakat the same time, binutils and make seem to build fine for mips15:06
methril_workwpwrak, i misss gta02-core project lately15:36
wpwrakmethril: you mean you didn't pay attention lately or that there was nothing to pay attention to lately ?15:38
methril_workwpwrak, i mean i didn't pay attention lately :)15:45
wpwrakmethril: ah well, it's been eerily silent for the last months. zero progress on the components, so i don't expect anything to happen there anymore.15:48
wpwrakmethril: i've been having fun with the ben, and everyone else stopped even before. so it doesn't look as if much would come out of this anymore :-(15:49
methril_workwpwrak, what do you think about the Beagle Hybrid project? started by pulster?15:51
methril_workand friends15:52
methril_workare you involved?15:52
wpwrakno, i'm not involved. dunno really. it's nice that they're managing all the technology, but i don't know how open it is15:54
methril_worki hope to get an "open" modern phone some day ;)15:56
methril_worki would bought a gta02-core, but unfortunately the project is not evolving anymore15:56
methril_workbut i'm more aligned with qi-hw, so i'll put my effort here too15:57
wpwrakqi-hw shows more promise of getting something done. less complexity and a more thorough approach towards openness.15:58
tuxbrain_awaymethril_work: the beagle-OM thing is promoted by Golden delicious computer not pulster16:09
methril_workups !! thank you tuxbrain_away16:09
methril_worki don't want to make confusion :)16:09
tuxbrain_awayand I agree with wpwrak, I bet one NN than qi will create a phone product earlier than this aproach, and by product not mean a hackish thing to play,  mean a final user product than also will be free, or at least the free as it can be16:11
tuxbrain_awaymaybe it will be simplier, but also be sure it will be orders cheaper16:12
tuxbrain_awaybtw happy to see you so active in qi now :)16:12
methril_worki'm coming back :)16:13
methril_worki'm happy to see your success ;)16:13
methril_workreally good work16:13
tuxbrain_awayheheeh success ??? I'nm still poor16:13
methril_workyou convince me to come to this community :)16:14
methril_workbut happy?16:14
methril_workwho is rich here?16:14
tuxbrain_awaysure I feel very very confortable here and I'm learning a lot16:14
methril_workthis is nice16:15
tuxbrain_awaynow I know conjures about SPI and bitbanging, also reading grimoires of microcontrolers.... and having a new master to learn .. archimagician werner16:16
Action: methril_work needs to learn a lot of conjures about Verilog HDL to help Milkymist 16:18
tuxbrain_awayyep, but I need to gain some experience points to jump level to read FPGA grimoires ,16:19
methril_workwell, you are getting a lot of experience points :)16:20
tuxbrain_awayyou know, pure marketing :)16:21
Action: methril_work also kneel down to the archimagician werner16:21
methril_worki don't think so, you are getting a better hw hacker ;)16:21
tuxbrain_awayok remember I redirect to your when I been asked for recomendation, I miss you man16:23
methril_worki miss you too16:23
tuxbrain_away(enter romatinc music here)16:23
tuxbrain_awayI will try to hack on SD after some changes on tuxbrain site (i hope this week will be enough) , then summon black forces of ATmega demons and NN registers16:27
methril_workare you going to port the arduino API to the BenNN?16:30
wpwrakgrmbl. defeated by my own genius. the program i wrote to transform mill coordinates cleverly calculates the clearance based on the maximum point in the toolpath. alas, in some cases, that's just the bottom, which can be fairly deep. and wham goes to mill hitting the piece at full speed. luckily, nothing broke.16:34
wpwraklarsc: i guess before too long, we'll get tuxbrain's breakout board kernel driver :)16:35
tuxbrainwpwrak: ETA for tuxbrain breakout board kernel driver: just after Xue release :P17:10
tuxbrainI don't want to eclipse any open project :P17:10
tuxbrainopen-> already open17:10
tuxbrainmethril_work: If by arduino API you mean program/compile/and flash an sketch trough NN this has been already achieved on debian and with serial port.17:12
tuxbrainbut need soldering the board and that is not elegant.17:12
methril_worktuxbrain, i was thinking of using the same sw in arduino and nanonote17:13
tuxbraingoal now is do it on jlime/openwrt trough SDIO17:13
methril_worklike a lib17:13
tuxbrainagain this was done17:14
methril_worklinking with a lib, and getting the same result that you could get with other Arduino boards17:14
methril_workok, then you could attract some Arduino hobbistics to NN!! ;)17:14
tuxbrainmmmmm but you should need at least some accecible gpio on NN17:15
tuxbrainwell more than the few actually avaible17:15
methril_worknot in Ben, what maybe in SIE17:16
tuxbrainDue I will need a micro to spi->uart , and I'm plenti of ATmega328 I will use that one for the fists fase17:16
tuxbrainyes SIE is more suitable to such goal17:16
tuxbrainbut step by step17:16
methril_workAtmega328 has USB port?17:17
tuxbrainand actually my sie is on other hands17:17
tuxbrainno but has SPI interface17:17
tuxbrainI will converte SDIO on SPI interface o NN17:17
tuxbrainon NN17:17
tuxbrainso NN will talk directly to the atmega17:18
tuxbrain(in theory)17:18
wpwrakhmm. 1.6 mm pcb into usb doesn't connect nicely. nominal thickness is about 2 mm and you notice the missing 0.4 mm. grmbl.17:22
wpwrakcan even be 2.36-2.92 mm. ah well, some plastic plus superglue then ...17:26
tuxbrainthen option a) make the own atmega powered breakout board an arduino it self, but due I will maintain same form factor as werner, I will sure go to option b) search for a smaller cheaper atmega compatible with 328 and with also spi/uart and do a SD to connect any 5V/3,3V module that has Uart interface... , maybe is there is some room on that uSD adapter we can enable as much as IOs possible of that small atmega and be able to to reflash it directly17:32
tuxbrain from NN to have a miniArduino like  in additon but this is a plus, and very far away on the schedule, but due we also can directly plug an arduino board to SD and communicatereprogram it through spi a with few discrete components as far as I can see by now, this has more chances to happen sooner17:32
methril_worktime to go home17:33
wpwraktuxbrain: you can just take an atmega4817:39
wpwrakif you want to add simple stuff, like some leds, connect a series of gpios to 0603 footprints.17:40
wpwrake.g., if you have gpio P1 through P3 and footprints F1 and F2, then F1.1 = P1, F1.2 = F2.1 = P2, F2.2 = P3   (Fx.y, x = footprint/component number, y = pin of that component)17:41
wpwrakbut it's probably better to keep things simple on that little board. you don't want to load it with lots of things. just increases the risk of someone breaking stuff.17:48
tuxbrainwpwrak ok agree, sorry I'm on mindstorm phase, having the serial functionality is far than enough, if people want a lot of gpios ready to be used though NN just buy an Arduino UNO , a simple breakout board and start play.18:15
tuxbrainor a SIE board :)18:15
wpwrakyou can always make board variants later. or maybe even sell pre-cut copper-clad boards. that way, people who know how to etch and solder can make their own.18:19
wpwrakand yes, there are even more profitable options :)18:20
tuxbrainbtw: as part of the mindstorm phase, I saw  that there is an realtive cheap arduino shield that is able to provide partial usb host capabilities though spi http://www.sparkfun.com/datasheets/DevTools/Arduino/USBHostShield-v12i.pdf, mmmm this should be easily extrapolated to the future sd spi nanonote interface isn't it?18:59
wpwrakheh, crazy chip:) yup, you could indeed use that one to implement a usb host19:02
tuxbrainmmmmm not sounds so a bad an USB->uSd interface :)19:04
wpwraknot trivial, thogh. you also need a boost converter to supply 5 V for VBUS19:04
wpwrakand i'd be worried about mechanical issues. with usb, you have quite some lever19:05
tuxbrainexternal source , and fexible cable19:06
wpwrakexternal source is messy19:06
wpwrakflexible cable ... hmm, maybe19:07
tuxbrainpluguin a ungry USB device to NN is crazy, is a 850mAh batt, you will need external source19:07
wpwrakif you have an external 5 V source it's most likely something USB-ish too. so you could feed the Ben from it19:08
wpwrakand the ben could then pass it on. only ~ 100 mA, of course19:09
wpwrakotherwise you need some fancy cable19:09
tuxbrainand also what will be the speed , sdio will reach 1Mbps though spi maybe as arduino?19:10
wpwrakyou can use the ben's clock and save the crystal. that's good.19:10
wpwrakyeah, something like 1 Mbps sounds reasonable. didn't really measure the timing19:11
tuxbrainI thing enough to 1.1USB then , good :)19:11
tuxbrainknowledge is cruel hungry bitch, more you feed it the more it wants, ok enough procrastinating, lets do some boring web manteinance19:13
wpwrakyou could design this such that you first plug the board on the cable/device, then insert it into the ben. make the case such that it extends to the surface on which the ben resides.19:13
tuxbrainwpwrak: cleaver :)19:14
wpwrakso you could use usb memory, wlan, bt, and such.19:14
tuxbrainwpwrak: after you finish wpan, work on this one and you will be not only a hero , a super fu&$ing greatest even hero on the NN community19:16
wpwrakunfortunately, the usb receptacle is too thick to fit under the board. so it has to be on top. it would go well above then ben's keyboard.19:16
rafatuxbrain: ha!!!.. he is already a hero I see... we need to find him super hero clothing ! :)19:18
wpwraklike this guy ? http://4.bp.blogspot.com/_aHPccFvfhGY/SRapr2kkCgI/AAAAAAAAAAo/GDwndDYeoUc/S220/obviousman.png19:20
rafayeah!.. but that is very light..19:22
rafahttp://upload.wikimedia.org/wikipedia/en/1/16/HardwareM0.jpg .. but you need some gym to be comfortable with that on you :P19:22
tuxbrainchange that knife for a beer, and the chain with a solder iron, and that what Imagine wpwrak19:24
tuxbrainthat is how I imaginte wpwrak19:24
tuxbrainor at lest his 15m bronze estatue19:25
wpwrakrafa: i notice that he uses quite a lot of metal to keep that belly in. so that may work :-)19:25
wpwraktuxbrain: ;-))19:25
rafatuxbrain: if we will use bronze then we need to hide it a bit here.. people used to steal every thing on the streets with bronze19:28
tuxbrainbronce or cooper?19:29
rafatuxbrain: bronce. btw, if you have visited Rome and you are okey how the people drives there, how they park the cars.. and how mess is the city.. then you will be comfortable when you visit buenos aires :).. it19:30
tuxbrainwell whatever, wpan= bronze, wpan+usb host=silver, wpan+usbhost+ethernet=gold :)19:31
rafawas very similar to buenos aires ( I have not visited many cities in europe, but from the few ones I did Rome was most like BA :D19:31
rafatuxbrain: good rank19:32
wpwrakwpan is actually the hardest of the three :) and i'm not sure you really want ethernet. you almost certainly have some kind of pc around anyway. so  2 x wpan.19:32
tuxbrainwpan+usbhost+ethernet+serial: platinum :P19:32
rafahaha :D19:33
wpwrak(gold, platinum, etc.) if you would send a few material sample my way, then i'd be more than willing to implement whatever crazy peripheral you can dream up :)19:34
tuxbrainethernet= an emergency micro server, network analizer, some ahem (I will not repeat this in publica again, ARP spoofing....)19:34
tuxbrainyou know is only matter of money not matter of will :P19:34
wpwraknaw, the emergency server doesn't make sense. totally different type of device.19:34
wpwraksame problem everywhere, i guess :)19:35
wpwraknetwork analyzer. dunno. the ben may be a bit fragile for this, particularly if you hang things off the uSD slot19:36
wpwrak(emergency server) a more sensible form factor would be a "brick". if you have lots of things you need to configure, maybe add display and touch screen.19:37
Action: tuxbrain is also thinking.... SPI allos bus, arduino also connects SD reader +ethernet to SPI, so we don't have to loose tha hability to read an sd on those crazy dongles :)19:38
tuxbrainwell let's stop19:38
wpwrakbut i think those wall wart type of small pcs that are popping up everywhere would be quite ideal for this kind of use19:38
wpwrakspi->arduino->uSD. boldly explore entirely new universes of slowness ;-)19:39
tuxbrainyes yes, nn is not for that porpuse, and once spi is integrated enough in the NN idiosicracy , a breakout board and good howto will fit for those who wants ethernet :P19:40
tuxbrainas our mater and mentor wolfi says:focus19:40
wpwrakyeah :)19:41
wpwrakben -> 3.3V/5V serial should be quite trivial19:42
wpwrakben -> USB host feasible, but significantly harder. more like in the order of wpan19:42
tuxbrainbite me , trivial for you f&%ng b$%&rd19:42
tuxbrainyou know how break  the spirit of a wannabe19:43
tuxbrainthat's why I will work on serial, and you will work on USB :P19:44
wpwrakalright, let's say "easy" :)19:44
tuxbrainand that's why people will end up puttinga usb-serial adaptor heeheheh19:46
wpwrakyeah, if you solve ben->usb,. you implicitly solve ben -> rs232 :)19:48
wpwrakwell, if that host is really good for that. not sure how far it can go19:48
wpwrakthose cute little spi-attached things from maxim often have nasty limitation19:49
wpwrakor wait, maybe that was microchip19:50
wpwrakanyway, that one may have nasty limitations as well19:50
tuxbrainwell is matter of try, put a breadboard, plug the maxim chip, plug NN and , start pluging devices :), in NN soft part? also drivers has to be implemented isn't it?19:56
tuxbrainwell time to sleep, more interesting dreams tonights, a giant boring web things TODO list chasing for me , meanwhile I hide on weak refuge mado of "wanna do" in land of procrastination20:01
tuxbrainmado ->made off20:01
tuxbraingood k820:01
wpwraktuxbrain: (sw) yes, needs a driver too, it seems20:05
kristianpaulwpwrak: are you implementing usb client capabillities of wpan in a fpga?20:08
Action: kristianpaul runs away20:08
wpwrakdoes this look like an fpga ? :) http://projects.qi-hardware.com/schhist/atusb/pdf_head/USB.pdf20:09
kristianpaulmay be a CPLD ;)20:10
kristianpaulwpwrak: just kidding was the only thing i got after read the backlog20:12
kristianpaulerror: src refspec master does not match any.20:27
kristianpaulerror: failed to push some refs to 'git@projects.qi-hardware.com:ben-gps-sdr.git'20:27
kristianpauli just followed indeferon instructions..20:27
kristianpaulWTF it seems it worked knw i can clone it..20:28
kristianpaulwolfspraul: http://paste.debian.net/93307/20:56
wolfspraulkristianpaul: and?20:58
kristianpaulwolfspraul: i cant push20:59
kristianpauli folled indefero instructions as you see20:59
kristianpaulsorry my only git experience was in github and i dint have this issue before21:00
wolfspraullooks like your local git config is messed up21:00
wolfspraultry with a new, empty directory21:00
wolfspraulmkdir ben-gps-sdr21:00
wolfspraulthe errors you get there are all local errors, I believe21:01
wolfspraulyou can disconnect from the network and I think you will still get those errors :-)21:01
kristianpaulerror: src refspec master does not match any.21:02
kristianpaulerror: failed to push some refs to 'git@projects.qi-hardware.com:ben-gps-sdr.git'21:02
kristianpaulwith new directory21:02
wolfspraulah you pushed something already?21:02
kristianpauli just was following indefero instruction to initialze the repo21:03
wolfspraulshould we just delete whatever was pushed to the server so far?21:04
wolfspraulyou may have run a lot of commands and it's harder to track down the source than to start over21:04
kristianpaulso delete project and start over again?21:05
wolfspraulyou tell me21:06
wolfsprauldid you commit something already?21:06
wolfspraulthe repo looks empty on the server21:06
kristianpaulno i dont21:07
wolfspraulI'll delete it on the server, then you try to commit21:07
wolfspraulstart with an empty dir, put a file inside, git add, git commit, git push21:07
wolfspraulok go ahead21:08
kristianpauldo i follow the first comming instruction that indefero point?21:09
wolfspraulwhich one?21:09
wolfspraulyes, just follow those instructions21:09
wolfspraulyes looks good21:10
kristianpaulahh now is working21:10
kristianpaulwas because i had an empty directory :P21:10
wolfspraulwe could have studied what went wrong in the old git (server) repo, but if you dont' mind we just delete it?21:10
wolfspraul(I only moved it aside for now)21:10
wolfsprauldo you want me to finally delete whatever we had there before?21:10
kristianpaulis empty21:11
kristianpaulsorry for the anoying errors21:13
wolfspraulaisa: you there?21:19
wolfspraulkristianpaul: no problem I like those small bugs21:20
aisaI am now.21:24
aisawolfspraul: I am now.  ;-)21:24
aisaSince my name is what got me here.21:24
wolfspraulwell I hope I don't disturb you21:25
wolfspraulabout the ben-nanonote branch21:25
aisaNo, I signed in to say hello.21:25
aisaplease, give me your feedback.21:26
wolfspraulI'm a 100% upstream guy, so I have no time to participate in such a branch21:26
aisaGood, this is very helpful to hear.21:26
wolfspraulif you want to do it, and especially if there are other who want to join, go ahead21:27
wolfspraulI would be careful though.21:27
aisaCan you be more specific about your warning?21:27
wolfspraulmaybe it's a good rule - if you join some new project, the first few weeks (months?), just follow what the others are doing21:27
wolfspraulwell the danger is that you will just create an abandoned branch21:27
wolfspraulyou are enthousiastic about it, make all sorts of modifications/customizations (that's what the branch is for)21:28
wolfspraulbut nobody spends time to merge it upstream21:28
wolfspraulhow is this sustainable?21:28
wolfspraulit can only be sustained by you21:28
aisaIndeed, I realize I'm in a position where my ideas may not be right,21:28
aisabecause of what you just said: I'm new to this project.21:28
wolfspraulthink this way: realize how small this project is21:29
wolfsprauland how many problems we tackle21:29
wolfspraulso focus is most important21:29
wolfspraulon the openwrt side, there are a few quite active people, for example kyak who's in this channel quite often21:29
wolfspraulwe should hear his advice too21:29
wolfspraulhe seemed to like your banner yesterday, and wanted more21:29
aisaI will likely need to hear this advice again, please don't be afraid to repeat yourself about it.21:29
wolfspraulkyak has been contributing great improvements into openwrt-xburst and openwrt-packages21:30
wolfspraulit's funny I look at that tracking_ branch in exactly the opposite way from you21:31
wolfspraulI try to find out what is different from upstream, and either get it upstream or delete it :-)21:31
aisaOh?  It would be helpful for me to hear how you perceive it.21:31
wolfspraulI wouldn't add one line of code to the diff21:32
wolfspraulaisa: what do you think?21:38
aisaI'm not sure yet.21:38
aisaI think to create the best experience possible on the NanoNote, we'll need to make changes that aren't appropriate for OpenWrt.21:39
aisaI'm willing to ignore this and favor your advice,21:39
aisaand work in the tree under your guidance until I've better understood qi-hardware.21:39
aisaYou did convince me to stay focused on OpenWrt rather than Jlime, last night.21:40
wolfsprauldo you have an example for a change that you think is not appropriate for OpenWrt?21:41
aisaYes, I have at least one more.21:42
aisaI'd like to modify /etc/config/fstab21:42
aisato include defaults for microSD devices.21:42
aisanamely, a vfat and ext2 microSD.21:42
aisaand then include various directories under /mnt for each type.21:42
wolfspraulyou mean to improve auto-mounting?21:42
wolfspraulgreat!!! I love it.21:42
wolfspraulyes that's the kind of small detail improvement we need :-)21:43
aisaThis is entirely what I have in mind at this point.21:43
aisarather than radical changes.21:43
wolfspraulnow the first attempt is to do it by using OpenWrt, rather than patching/modifying it21:43
aisaMy thinking was this:21:43
aisaIt is very obvious how to do it as a patch to OpenWrt.21:44
wolfspraulif that is truly impossible, we should think about the smallest possible modification (enhancement) to OpenWrt, maybe an additional hook, or whatever, and propose that upstream as well21:44
aisait is not immediately obvious how to do it without a patch.21:44
wolfspraul(that's much later)21:44
aisahowever, I hoped that as I worked more with OpenWrt, I would be able to discover new ways of making old ideas work.21:44
wolfspraulwell you looked into the tech details of this more than me, so you know better21:44
aisaand revisit some of the patches and convert them to things that can go upstream.21:44
wolfspraulthat's what we have the tracking_ branches for21:44
wolfspraulit's exactly that 'upstream or delete' staging area21:45
aisaOk then.21:45
wolfspraulso we can move forward, make customizations/improvements today, and fast21:45
wolfspraulbut we keep in mind that we do not have the resources to maintain an entire OS, so the pressure in the tracking_ branches is to get stuff upstream or to delete it21:45
wolfspraulsmall things like your banner of course should stay21:46
wpwrakthis sort of integration changes is just what's the problem of distros that try to craft the entire rootfs. they force you to do everything inside their build environment. kinda as if you'd have to put your main() into libc in c programs.21:46
wolfspraulok fair enough. but hopefully the system is flexible enough to allow for easy customization. that's what it's all about I'd say.21:46
aisaAh, I have one other problem right now that I want to fix.21:47
aisathe permissions of the files in data/qi_lb60/files21:47
aisaare the person building, rather than root,21:47
wpwraka packet-based approach just avoids the issue by leaving all the "glue" to the thing combining stuff. now, one question would be how to go from rootfs-making to package-making21:47
aisaor better, rather than being specified.21:47
wolfspraulpermissions in git, or permissions in the rootfs?21:47
aisathe permissions in the rootfs were the permissions in git.21:48
aisaunless I did something wrong, of course.21:48
wolfspraulok (I meant you are talking about problems accessing these files in git)21:48
aisano, I can access them fine.21:48
wolfspraulgo ahead what about the permissions?21:48
aisabut they don't go into rootfs correctly.21:48
aisathere does not seem to be any step to say "this file should be mode 775"  this file should be owned by root, etc.21:48
aisathey're just copied in as-is.21:49
wolfspraulhe. ok. I would need to look through the sources, but it sounds like it's a meaningful feature and OpenWrt should have a way to specify owner and permissions in the dest rootfs21:51
wolfspraulmaybe for now we can add a hook somewhere, some sort of 'rootfs post-processing'?21:52
wolfspraulthen bring it up on the OpenWrt side?21:52
aisaI was thinking we could create a package, like base-files,21:52
aisabut call it nanonote-files or something,21:52
aisaand move this stuff into there.21:52
wolfsprauleverytime I see 'nanonote' or 'qi' I cringe :-)21:53
wolfsprauljust fyi21:53
wolfspraulbecause I'm the one deleting all this stuff 2 years later...21:53
aisai can't call it base-files,21:53
wolfspraulsure sure, I just give you a brain dump21:53
aisaand I can't add it to the base-files package.21:53
wpwrakwolfspraul: btw, regarding jlime, if they make a variant that doesn't have the offending codecs, would that be sufficient for making it safe enough to link to them or would you also need them to have no packages with the offending codecs at all ?21:53
wolfspraulyou should feel free21:53
aisaso you give me a better idea or I'm making your cringe ;-)21:53
wolfspraulone time someone patched the joe package and called it 'qijoe' (it's still called like that)21:53
aisaoh, right.  I did see that I guess.21:54
wolfspraulI imagined a horde of hundreds of qi... packages, all broken and unmaintained of course :-)21:54
aisaof course.  :-)21:54
aisaoh, that reminds me,21:54
aisathere is a testing page on the wiki...21:54
aisalike "things to do with a new image."21:54
wolfspraulreally? we have a testplan?21:54
aisanot really, I need to improve it.21:54
wolfspraulaisa: what about my proposal of a rootfs post-processing script?21:54
aisaI think it duplicates package functionality.21:55
wolfspraullike I said I look for ways to be as non-intrusive on the OpenWrt side as possible21:55
aisaHowever, do you remember the problem with including gforth in base?21:55
aisarather than it being a package?21:55
aisathe gforth package needs a post-install script to run.21:55
aisaso if we went with your idea,21:55
wolfspraulI would also ask OpenWrt (you sent a pretty nice mail there, let's see whether someone answers, they first have to get to know you...)21:55
aisait has to accept post-install scripts from other places.21:55
aisaThat was the result of our conversation yesterday.21:56
aisaIf we want to push upstream to openwrt,21:56
wolfspraulwpwrak: no packages, no instructions21:56
aisaI need to develop a relationship with upstream.21:56
wolfspraulthink from their perspective, then it should be clear. they are targeting a company (infringer).21:56
wolfspraulone defense strategy for that company is "it's not us"21:57
wolfspraulbut there needs to be a legally strong enough boundary between that company and whoever is infringing21:57
wolfspraulotherwise the company will not get away with that excuse/explanation21:57
wolfspraulwhen the patent lawyer can make a legally usable link between you and the infringement, you are done21:57
wolfspraulso it starts with Tuxbrain21:58
wpwrakwolfspraul: yes, my question is about what degree of distance is sufficient21:58
wolfspraulof course stuff that tuxbrain sells, and how it behaves out of the box21:58
wolfspraulthen tuxbrain.com21:58
wolfspraulanything hosted there21:58
wolfspraullinks to other websites (!)21:58
wolfspraulyeah well. they say "tuxbrain, you infringe"21:58
wpwrakwolfspraul: eventually, it's wget and gcc :)21:58
wolfspraultuxbrain says "no, it's not me. how did you find this?"21:58
wolfspraulyes true21:58
wolfspraulbut that is too cryptic, no judge would follow21:59
wolfsprauland 'they' (the patent holders) wouldn't care either21:59
wolfspraulthey do want to make sure that regular people are out, they are the masses anyway and the reason the big corps are paying21:59
wolfspraulsome fringe hackers will always make it work, nobody cares21:59
wolfspraulthat's why I say, before you think about 'packages', think about 'instructions' on a website, and links22:00
wolfspraulthat's how it starts in their logic22:00
wolfspraulthen there may be an 'installer'22:00
wolfsprauldownload file, install it22:00
wolfspraulnot good22:00
wolfspraulyou link to that?22:00
wolfspraulnot good22:00
wolfspraulif there are sources, and you need a very elaborate 100 steps process with cross-compiler and what not, I think you are safe22:00
wpwrakso openwrt has no patent-infringing packages at all ?22:01
wolfspraulopenwrt.org has22:01
wolfspraulso there should be no link from qi-hardware.com to openwrt.org22:01
wpwrakbut there's no link from qi-hw to openwrt.org ?22:01
wolfspraulof course I would think there are22:01
wolfspraulone might even argue using the word 'openwrt' is already too much22:01
aisalibrewrt :-)22:02
aisawell, except they're apparently going to have to change it.22:02
wolfspraulI could give you a URL that directly points to mp3 capable opkg files, built upstream...22:02
wolfspraulso definitely, any link to openwrt.org is dangerous22:02
wolfsprauljust a few more clicks and you have the opkg22:02
wolfsprauland then a few more clicks (well, we have no good installer right now), and you play an mp322:02
wolfspraulsisvel for example has employed technicians/engineers, who try to come up with a way to demonstrate that someone is infringing22:03
wpwrakso jlime could just not build an xburst mp3 package ?22:03
wolfspraulthe lawyers are waiting for these guys to show them the steps22:03
wolfspraulthen the lawyers decide whether they have a case22:03
wolfspraulthese tech guys of course try to find the _easiest_ steps for their lawyers22:04
wolfspraulkeep in mind, for the FreeRunner for example, it was really easy :-)22:04
wolfspraulthe capability was just present right in the out-of-the-box rootfs22:04
wolfspraulon the NanoNote, we have no easily usable installer22:04
wpwraksure. that was bad22:04
wolfspraulyou have to setup a usb/eth interface22:05
wpwrakso that's why you like openwrt :)22:05
wolfspraulyou have to copy the file with scp22:05
wolfsprauloh well, if aisa improves auto-mounting maybe microSD would be easier22:05
wolfspraulmaybe he even adds good fat32 support, then they could use a Windows host22:05
aisafat32 to uboot?22:05
wolfspraulthen you need to run 2 commands22:06
wolfspraulopkg install, and then some console command to play an mp322:06
wolfspraulnot sure whether console alone would be a turn-off nowadays22:06
kristianpauljuan64bits: hey, there, u awake?22:06
wolfspraulwpwrak: you can never be 100% sure whether you are safe or not22:06
wolfspraulthe key is for the seller in a patent infected country to not 'infringe'22:07
wolfspraulhim (!) - he must not infringe22:07
wolfspraulothers can infringe22:07
wolfspraulthat's the point22:07
wolfspraulif the end user infringes - no problem they will not go after them22:07
wolfspraulbut the seller (rather importer) must not be shown to infringe, otherwise they will come down on him22:07
wolfspraulmakes sense?22:07
wpwrakbut he can't really not be sure that he doesn't infringe either22:08
kristianpauljuan64bits: is it normal start SIE witht the ADC program showing ramdon values and then got no display refresh from data in A0 or A122:08
wolfspraulwe could make total no-brand devices22:08
wolfsprauljust all black22:08
wpwrakthat wouldn't solve the problem22:08
wpwraktuxbrain imports them, apparently for some purpose22:08
wolfspraulif the seller does not infringe they have no case22:09
wpwrakif a judge is willing to follow that line of reasoning, they still get him22:09
aisawolfspraul: I'm heading out soon.  Thank you so much for your feedback, I appreciate you coaching me very much.22:09
kristianpauljuan64bits: all was working fine i just restart the board now i the display is like craxy showing data, with no reflesh iw, if in wire analog inputs to GND..22:09
wolfspraulah, thanks. you also make me think about the best way forward. it's great to have you here!22:09
wolfspraulwell they will never openly explain to you who they go after or not22:09
wolfspraulI'm just explaining that 'infringing' thing22:10
kristianpauljuan64bits: hmm but it was responding, and please dont respond me in PM !!!22:10
wolfspraulwho is infringing? that's the key legal question22:10
kristianpauljuan64bits: or you want a PM chatting? :)22:10
kristianpauljuan64bits: i'll try but it is supposed to be done at startup isnt?22:10
wolfspraulso let's bring in the real world22:11
wolfspraulour sales numbers are very very small22:11
wpwrakas long as any level of indirection is considered as not changing things, there's no way to be safe22:11
wolfspraulnot 'any level'22:11
kristianpauljuan64bits: and it was working just i restarted and is gone22:11
wolfspraulI explained it to you.22:11
wolfspraulit must be legally strong enough22:11
wolfspraulthey are lawyers, not engineers22:11
kristianpauljuan64bits: fpga isn't configured, make sense, thanks22:11
wolfspraulgui installer22:11
wpwrakwhich basically depends on the judge's mood22:11
kristianpauljuan64bits: sure you can do public irc chatting with pidgin :)22:12
wolfspraulor what the patent holders know will work, will work as 'evidence' for the infringement22:12
wolfspraulthe steps they are outlining is evidence22:12
wolfspraulwith screenshots etc.22:12
wolfspraulclick here, click there, bla bla bla22:12
wpwrakyes, there are obvious pitfalls. clear instructions. package part of the "official" set hosted at qi-hw or tuxbrain, whatever22:12
wolfsprauldon't think theoretical, or like an engineer22:13
wolfspraulif you want to keep the actual risk low, you need to think about how easy it is with your device to infringe22:13
wolfspraulthe easier it is, the higher your risk that they will single you out22:13
wpwrakit's less clear to me if, say, you have the official repository, sans offending packages, e.g. hosted at qi-hw22:13
wolfspraulright now we are pretty safe22:13
wolfspraulbecause the technology is still hard to use22:13
wpwrakplus a link giving credit to jlime (jlime top-level page)22:13
wpwrakand then jlime have a set of packages as well, including the offending one22:14
wolfspraulunfortunately they only look at your device from one perspective: "what is the easiest way to get an mp3 to play here"22:14
wolfspraulthen they need to link those steps to you22:14
wpwrakso by that top-level link, do you assume responsibility for anything jlime does ?22:14
wolfspraulso if they use google and download an opkg from a russian server (sorry for our russian friends), they have no case22:14
wpwrakunless you document the google instructions22:15
wolfspraulbecause your lawyer would say "that's not us", and their case (that you are infringing) is no case22:15
wolfspraulso a wiki or forum is borderline22:15
wolfspraula wiki for example needs to be 'governed/policed' in a reasonable way22:15
wolfspraulwhat that exactly is is a bit hard to define22:15
wolfspraulagain: risk22:16
wpwrakyup. if the complain, you certainly have to remove instrucitons or you're in trouble22:16
wpwrakanythig beyond that, gray zone22:16
wolfspraulif you have a big fat link on your homepage, the risk is that they will fire up their laser printer, and make a nice printout, and add it to the evidence22:16
wpwraksure. obvious trap22:16
wolfsprauloh for sure22:16
wolfspraulafter they notify you, you definitely have to take it down22:16
wolfspraulthat would be quite gentleman-like actually22:16
wpwraknow, how about my example ? you have a qi-hw repository of "safe" jlime packages. by default, the package installer only goes there22:16
wolfspraulthey could also just collect evidence and hammer down on you one day22:17
wolfspraulyes exactly, that's what we have with openwrt22:17
wpwrakbut jlime have do whatever they like to do on their side, and you merely provide a proper modern-day reference, i.e., a link to jlime (as jlime, not anywhere inside)22:17
wpwrakwould that sufficient separation to make you feel safe ?22:18
wpwrakor safe enough :)22:18
wolfspraulif I would be the importer, I would feel safe as long as there is no <a href> link (let's call it 'active link')22:18
wolfspraulI would not worry about the word 'Jlime'22:19
wolfspraulor 'OpenWrt'22:19
wolfspraulbut I am not the importer :-)22:19
wpwrakbut no link to <a href="http://www.jlime.com/">Jlime</A>22:19
wolfspraulI think the moment they go via a search engine, or type a url into the browser, they cannot link the infringement to you anymore22:20
wolfspraulI would feel safe without such links, yes.22:20
wolfspraulthose links carry strong legal value in many Western countries nowadays22:20
wolfsprauldon't ask me, but hey, they understood the concept of 'link' and they think it's a big thing22:20
wpwrakyes, unfortunately22:20
kristianpauljuan64bits: public chat ok then?22:21
kristianpauljuan64bits: so i think is a bug what i pointed cause right now ADC is reading22:21
wpwraki'm just not sure how many levels of redirection would be required there22:21
wolfsprauljuan64bits: yeah! :-) don't be worried22:21
wolfspraulit's a chaotic place anyway here. you can even write in Spanish if you like.22:21
kristianpaulso something FPGA is not programed22:21
kristianpaulsi español !22:21
wolfspraulwpwrak: it's not about 'levels of indirection'22:21
wolfspraulturn off your engineering mind for a second22:22
wolfspraulfor them it's a sequence of steps22:22
wolfsprauland they need to show that it's always _YOU_ behind those steps22:22
wpwrakwell, indirections = steps22:22
wolfspraulthen they can show that _YOU_ are infringing22:22
wolfspraulthe moment there is even just one step that is not you, they have no case against you22:22
wpwraki don't think they have to show it's you all the time22:22
kristianpauljuan64bits: are you aware of SIE file system generation?22:23
wpwrakthere's that concept that, by linking somewhere, you're approproating the material there22:23
wolfspraulthat's the best I know22:23
wolfsprauloh sure22:23
wolfspraulyou set the link!22:23
wolfspraulof course that is you22:23
kristianpaulany way i just found this and want to tell somebody  i guess near or more SIE's22:23
wpwrakeven if you're not behind the material or even if it didn't exist when you set the link22:23
wolfspraulif you set the link, you become responsible for what is going on on the other side22:23
wolfspraulotherwise you need to remove the link22:24
wolfspraulsimilar like wiki, forum22:24
kristianpaulnow i need wire this MSPS ADC, cause i need or buy an scope 10Mhz or use SIE as one :)22:24
wpwrakso if you add a link to google ... :)22:24
juan64bitskristianpaul wait me a few minutes....22:24
wpwrakor www.yahoo.com ...22:24
kristianpauljuan64bits: i can for sure :)22:24
wolfspraulfirst I doubt many sites have a link to google, and if they do, I think the (separate) concept of a search engine is understood by every lawyer and judge22:24
wpwrakafter all, there are yahoo groups and such. chnces are you can find a path that you can just click though without ever entering any text22:25
wolfspraulso then you type 'run mp3 on nanonote'22:25
wolfsprauland then you find a link pointing to which server?22:25
wolfspraulif it's not your server, again, you are not infringing22:25
wolfspraullet's stay in reality22:25
wpwrakbut then the link to jlime.com or openwrt.org would be okay as well22:25
kristianpauljust tell google dont point to us :)22:26
wolfspraulthey try to proove that you are infringing22:26
wolfsprauland normal people need to understand them/believe them22:26
wpwrakas long as they don't put the "bad" instructions right on their front page or such22:26
wolfspraulwhat is the difference between that one link or not22:26
wolfspraulin the end, you cannot (in reality) have mp3 functionality22:26
wolfspraulthat's the point22:26
wolfspraulall your normal customers will not be able to play mp322:26
wolfspraulyou can dance around the legal definitions of 'links' as long as you like22:27
wolfspraulthe end result will be:22:27
juan64bitsI'm back22:27
wolfspraul1. you cannot use mp3 in marketing22:27
wolfspraul2. your normal customers will not be able to play mp3, neither out of the box, nor after some installer easy enough for 99% of people to use22:27
wolfspraul3. people such as those in this channel will always be able to run whatever software they like. to them it also doesn't matter which link is where, or how direct the link is.22:28
wpwrakyes, good so far22:28
kristianpauljuan64bits so as i said i think there is a bug cause something SIE FPGA is not programed22:29
wolfspraulok that's all I have :-)22:29
wolfspraulno mp3 out of the box22:29
kristianpaul(talking about the ADC demo that came by default it seems)22:29
wolfspraulno mp3 installable in ways that would proove that the importer is behind it/endorsing it/linking to it22:29
kristianpauljuan64bits: the other question was, how do you generate the root file system for the board?22:30
wolfspraulseparate package repository22:30
juan64bitsyes, maybe , but some times is a problem in the board22:30
wpwrakendorsing is tricky22:30
kristianpauljuan64bits: are you using openwrt to include in the root ubi image, or just mnanual hacking?22:30
wpwrake.g., what if you paid jlime to make a set of mp3-free packages for you ?22:30
wpwrakbut in their spare time, they still maintain their "full" repository on their own machines22:31
juan64bitsthe fs is the same of the nanonote, with manual hacking.22:31
wolfspraulremove anything that infringes from collaborative forums, wikis, etc. either in advance, or if you like the thrill, after notification22:31
kristianpauljuan64bits: also is tooo bored i cant to MSPS measures in SIE and so far i was not able to find a TI MSPS ADC chip compatible pin with to replace the current TLV154822:32
juan64bitsalso with ubi using the .tar.gz and adding the files compiled for SIE22:32
wolfsprauljuan64bits: what 'problem in the board' do you mean?22:32
kristianpauljuan64bits: the _only_ thing i need from SIE now was and scope, but seems i'll end adding it with and external board, but i guess SIE was not planed to measure Mhz data out of the box ;) so np22:33
juan64bitssometimes i have to touch the jtag resistors22:33
kristianpaulhmm why?22:34
wolfspraulkristianpaul: Adam ran all tests on all boards, so it should work, at least to the degree as provided by the tests we ran.22:34
kristianpaulwolfspraul: yes it works, waht i pointed seems more a sofware issue, so dont worry :)22:34
juan64bitsyeah, Mhz wasn't planed22:35
wolfspraulok, good22:35
kristianpauljuan64bits: right now i got  a TLV1562 i hope do the job :=22:35
wolfspraul(that it's probably a sw issue...)22:35
wolfsprauljuan64bits: we need someone to push those 'manual changes' into a clean OpenWrt target22:36
wolfspraulwpwrak: you wouldn't want them to be able to proove that you are helping someone to infringe22:38
wolfspraulhelping someone to commit a crime is better than committing the crime yourself, but it will still get you into trouble22:38
wolfspraulespecially if you have profit motives behind it etc.22:38
wolfspraulbut I think even in the context of sisvel this kind of case would never be brought forward22:38
wolfspraulI want to stay real, not all just in theory.22:39
wolfspraulthe most aggressive enforcer I know is sisvel (mp3)22:39
juan64bitsok, but those "manual changes" maybe just creating new feeds for openwrt22:39
wolfsprauljuan64bits: we need a clean target22:39
wolfsprauleven if it's just a 'few' simple steps22:39
kristianpaulor documentation in the wiki22:40
wolfspraulwpwrak: after that, mpeg la (mpeg4/h.264)22:40
kristianpaul(i think)22:40
wpwrakalright, so you're saying that you think it would be safe, to, day pay jlime for maintaining the "good" packages and providing them on a qi-hw server22:40
wolfspraulafter that, I don't even know already. I think the rest are all people going after big money, and then it's all fair game anyway.22:40
wpwrakincluding making sure the package installer doens't by default stray elsewhere22:40
wpwrak(big money) yeah, like RIM22:41
wolfspraulyes sure that's what we do with OpenWrt22:41
wolfspraulI'm pretty sure David/Tuxbrain would accept this.22:41
juan64bitsi think in the "feeds" beacuse we are using the minimal configuration of NN for get our fs, and then the applications for SIE are added22:42
wpwrakalright. even if jlime/openwrt have their own "bad" repository, that then shouldn't be high-risk22:42
wolfspraulwpwrak: that's what I would think, yes22:43
wpwrakjust like you'll probably not get charged for helping a murder if your baker buys a gun with your money and shoots his wife. probably not even if he's been talking about wanting to kill her.22:43
wpwrakokay, good22:43
wolfspraulalthough I would remove links there (<a href> links)22:43
wolfspraulI cannot imagine that sisvel would bring forward a complicate case that involves the inner operations of big free software projects like OpenWrt or Jlime/OE22:44
wpwraklinks to openwrt/jlime-the-project, not only links to offending packages/instructions22:44
wolfspraulyes all <a href> links22:44
wolfspraulif you operate in this way, I think it will be very hard for them to show that you are the infringer22:44
wolfspraulat least that's a good 'base' to operate from22:44
wpwrakoverly constructed cases can also work against the attacker. not only may they lose (which they probably don't care about), but they'll also look weak in public22:45
wolfspraulyou can never rule out that they come after you anyway, but with my experience I would feel safe in this way, yes22:45
wolfspraulyes correct22:45
wolfspraulno they know that22:45
wolfspraulthey don't want to stir up a lot of dust22:45
wpwrakrafa: did you read all this ? :-)22:45
wolfspraulthey want to make money22:45
wolfsprauland - by the way22:45
wolfspraulit's easy to point at the bad sisvel guys22:45
wolfspraulafaik it's actually Phillips22:45
wolfspraulthey just created this no-name outfit so that the bad reputation won't fall on their 'good' brand22:46
wpwrakPhilips ? the dutch company ?22:46
wolfspraulthe way it works is that you create a no-name company, Sisvel here22:46
wolfspraulthen you 'sell' the MP3 rights to that company22:46
wpwrakoh, lovely22:46
wolfspraulin this case I think for 200-400 million USD22:46
wpwrakso no nxp parts :)22:46
wolfspraulso Sisvel starts with a huge amount of debt, in their books22:46
wolfspraulof course they get a credit line because they are backed by Phillips22:47
wolfspraulbut legally it's now a separate company, with lots of debt22:47
wolfsprauland the poor lawyers have to work down the debt mountain22:47
wolfsprauland take the blame :-)22:47
wolfspraulmeanwhile on Phillips side all is bright & sunny22:47
wpwraksomething useful to mention in press releases when that sort of lawsuit hits you22:48
wpwrakphilips are indeed concerned about such things22:48
wpwrak(one L, if we're talking about the dutch guys)22:48
wolfspraulsure but they would never talk about it22:48
wolfsprauland those 'links' (he he) would be very hard to demonstrate22:49
wolfspraulyou would need to be able to get copies of the sisvel shareholder/founding documents22:49
wolfspraultheir coporate charter etc.22:49
wpwrakit should be easy to prove the origin of the IP22:49
wpwraki mean, that's the first thing you'd challenge22:49
wolfspraulI'm not saying the lawyers are all great, but well if a poor little lawyer is employed there and presented with this mountain of debt, what shall he do :-)22:49
wolfspraultry to even tell them which patents they enforce...22:50
wolfspraulsorry typo "try to even get them to tell you..."22:50
wpwrakno, i mean when they hit you with a lawsuit. the they can't hide these things anymore22:50
wolfsprauland then, whoever filed the patent, what does that proove? the rights could have been sold or licensed many times22:50
wolfspraulall those agreements are private22:50
wpwrakso after the pulster incident, you should have all the information22:50
wolfspraulbut they may slap confidentiality agreements all over you anyway22:51
wolfspraulask Pulster22:51
wolfspraulI have seen some small parts.22:51
wpwrakhmm, i doubt they could require this if the thing makes it to court22:51
wpwrakclaiming they don't have the rights to sue you it pretty much standard practice. then they have to put something on the table.22:52
wolfspraulmaybe Pulster knows more, I don't22:53
wpwrakdid pulster go to court ? or was it just preliminary injunction and then settled out of court ?22:53
wolfspraulI'm just describing what I believe is a safe strategy to avoid patent infringement, in the cases like mp3/mpeg4 etc. where we are at least dealing with someone reasonable/predictable patent holders22:53
wolfsprauloh no22:53
wolfspraulbig big case22:53
wolfspraultens of thousands of EUR cost22:54
wolfsprauland so on22:54
wpwrakso it did go to court ?22:54
wolfspraulyes I think it did22:54
wpwrakwith a verdict ?22:54
wolfspraulPulster doesn't talk about it much, maybe Harald knows more.22:55
wolfspraulhe is selling NanoNotes though :-)22:55
wpwrakah, good :)22:55
wolfspraulfrom all I know, the patent problem right now really comes down to sisvel and mpeg la22:56
wolfspraulafter that there are endless trivial patents, but I think they are enforced not very systematically22:56
wolfspraulmaybe if you are in a very specialized field22:56
wolfspraullike Elphel is very very concerned about bundling stitching software with their panoramic cameras22:56
wpwrakseems that rafa isn't around at the moment. well, i think we have a proper delineation between dos and dont's now.22:56
wolfspraulthere seem to be some specialized 'stitching software' enforcers22:57
wolfspraulso Elphel will only sell cameras (even panoramic ones) that 'cannot' do stitching22:57
wpwrakluckily most of the trolls only seem to smell blood if you're making millions22:57
wolfspraulsame kind of chinese wall as I described earlier22:57
wolfspraulyes correct22:57
wolfspraulthe more money you make the bigger the patent problems becomes22:57
wpwrakapropos chinese, do you know why the MIPS patent expiration didn't work ?22:58
wolfspraulthat's why I say to my knowledge only sisvel and mpeg la are systematically enforcing22:58
wolfspraulI don't know22:58
wolfspraulthey must have made a proposal that the other side couldn't refuse22:58
wolfspraulif we are lucky, sisvel is history in a few years22:59
wolfspraulthat would be great22:59
wolfspraulmy feeling is it's a one-time shop setup to milk the mp3 patents to death, work down the debt mountain they were started with etc.22:59
wolfspraulbut then, you have a functioning organization22:59
wolfspraulyou could fill new IP into this pot :-)23:00
wpwrakgo to sisvel.com, click on contacts "SISVEL: AN IDEAL IP PARTNER FOR YOUR ORGANIZATION. CONTACT US TODAY."23:02
wpwrakseems that they're already looking for material :)23:02
wolfspraulyou do find the cruel stuff, don't you?23:02
wpwrak"news & events", DVB-T2, UHF RFID, Stereoscopic 3D, LTE, ...23:03
wolfspraulin their History they say they got the MP3 patents from Philipps23:05
wpwrakyeah, that's all they write about besides autio MPEG. nothing to worry about for now.23:05
wolfspraulso who is the bad guy now?23:05
wolfspraulwhat does Philipps think someone else will do after they sell the IP for xxx million USD? :-)23:05
wolfspraulbut well that's how it works... so Philipps can continue talking about being the 'good guys'23:06
wolfspraulmaybe sponsor a free software conference?23:07
wolfspraulat least they have enough money for free drinks and stuff23:07
wpwrakif the philips (one L, one P at a time) patent is used, one could perhaps innocently mention that in press releases :)23:07
wolfspraul"The company was so successful in managing the Indesit patents that major international players in the consumer electronics industry, such as Philips, France Télécom, TDF, and IRT, turned to SISVEL to license some of their most valuable intellectual property, including their patents relating to the MPEG Audio standard (Layers 1, 2, and 3)."23:08
wolfspraulI think Sisvel paid 200-400 mln USD23:08
wolfspraula lot of money23:08
wolfspraulneed to make that back first23:08
wolfspraulthat's their focus23:08
wolfspraulthey are very very focused from my experience with them23:09
wpwrakwhat do you mean with "very focused' ?23:09
wolfspraulfocused on making money23:09
wolfspraulof course they don't care how much damage they create on your side23:10
wolfspraulthey are focused on their money23:10
wpwrakah, that's a relatively popular obsession23:10
wpwrakdamage is an externality23:10
wolfspraulthat's why I dont' think they will go after cases that enter grey/questionable legal territory23:10
wolfspraulthey don't want to bind their resources on such nonsense23:10
wolfspraulit's a simple machine, quite clean actually23:10
wpwrakyeah, or weaken their attack on worthless targets23:10
wolfspraulthey show that you are infringing, they bring forward the case23:11
wolfspraulof course they know how to use the law to the last little comma23:11
wolfspraulwhich is open to you too, of course, only that you don't have a pool of 100+ full-time lawyers, with tons of prepared templates, prior cases, etc.23:12
wolfspraul"SISVEL's management of the MPEG Audio licensing program is a prime example of its role in protecting intellectual property rights and highlights the companys transformation into a global player and leader in the licensing community."23:15
wolfspraulouch, painful23:15
wolfspraulmaybe we should point them away from our field somehow23:15
wolfspraulmaybe some patent portfolio in oil & gas exploration?23:15
wolfspraulI hear that's a quite cash rich industry as well.23:16
wpwrakin illustrative techniques for religious figures ? maybe then some fundamentalist will take care of them for us :)23:16
rafawpwrak: if I read? from where? :)23:30
wpwrakrafa: welcome back ! :)23:30
wpwrakrafa: i asked wolfgang about how jlime offending packages could be integrated better into qi-hw23:31
wpwrakrafa: the bottom line is that the problem packages shouldn't be installed, easy to install, or that there should be instructions on qi-hw for installing them23:32
wpwrakrafa: also, if jlime hosts a any of the problems above (package, instructions, etc.), then qi-hw shouldn't have a link to that material or to jlime.com23:33
wolfspraulwpwrak: (not to discredit your great summary) I believe rafa already knows this.23:33
wolfspraulat least I was under the assumption that he does23:33
wpwrakrafa: so if jlime can make a package set and put it on some space at qi-hw, that would solve most of the issues23:34
wolfspraulwe can create a chinese wall separating infringing and non-infringing technology anywhere we like, it just has to be clear/clean enough23:34
wpwrakwolfspraul: i'm not sure all the details are clear23:34
wolfspraulthe current proposal is to see tuxbrain.com/qi-hardware.com/sharism.cc as one system23:35
rafawpwrak: wolfspraul : yes, I was guessing that.. but deeper.. the jlime/debian image at qi-hw sites.. could those have links to repositories of packages or web sites?..23:35
wolfsprauland to have no infringing stuff inside those walls23:35
rafa(web sites= links to jlime.com which could have packages with offensive software)23:35
wpwrakbtw, how about links into jlime.com, to  material unrelated to the problem pakcages ? (e.g., other packages or maybe also instructions)23:35
wolfspraulrafa: no, links are not good23:35
wolfspraulI walked wpwrak through the entire legal logic before.23:36
rafawolfspraul: yes, I guessed that as well..23:36
wolfspraulit starts with the infringer, and the other side trying to proove that it is in fact that person/company who is infringing, or helping someone else infringe.23:36
wolfspraulif that person/company can make the point that it is 'not them', the other side has no case23:36
wolfspraulso you need to start thinking from tuxbrain's side23:37
rafawolfspraul: so finally we can have something. I guessed this: an jlime image without offensive software, without links to sites explaining the bad things (bad things=for example, how to install mp3 software) and23:37
wolfspraulhow can tuxbrain claim that he is not infringing?23:37
rafawithout links to repositorirse packages with that kind of software23:37
wolfspraulyes, then I believe Tuxbrain could import or sell such devices and be reasonably safe from an infringement claim from Sisvel23:37
wpwrakyes, but there's a big gray zone. e.g., links to the other site. now, if someone has a problem with, say, vi, and there is an answer on the forum at jlime.com, would a link to that answer on qi-hw be okay, or would that still count as an posible infingement ?23:38
wolfspraulthe less links to sites that host infringing technology the better. but if the link is about a totally separate topic, I wouldn't know how they could use that link in a series of steps to proove infringement.23:39
wolfspraulthey are trying to proove an infringement of MP3 patents!23:39
wolfspraulnot come up with a random series of keypresses what will let you play MP323:39
wpwrakgood. i would consider this too far-fetched as well. but then, of course, IANAL23:39
wolfspraulSisvel is relatively predictable23:40
wolfspraulthey just want to tax mp3, done23:40
wolfspraulthey have no educational mission, or any mission beyond making as much money from the mp3 patents as possible23:40
wpwraki guess it would depend a lot on whether someone somewhere manages to make such a connection stick. in that case, they may want to give it a try23:40
wolfspraulI doubt that.23:40
wolfspraulFreeRunner had mp3 in nand :-)23:41
wolfsprauleven with our cryptic installers we should be safe23:41
wolfspraulbut I wouldn't take that chance23:41
wpwrakwell, freerunner ran into problems :) qed23:41
wolfspraulinstructions in the wiki are a gray zone, the question is how many resources you have to devote on policing23:41
wolfspraulbut you definitely cannot just leave links or instructions there, especially if you know they are there23:42
wolfspraulthey do understand the concept of 'wiki'23:42
wpwrakokay, so links to unrelated material are okay too. just not "top-level" of the offensive material23:42
wolfspraulthey are not rogue23:42
wolfspraulthey would not use one of their own people to insert infringing stuff into your wiki, then drag you out, etc.23:42
wpwrakhah, who knows :)23:43
wolfspraulif it's really unrelated, I wouldn't know how it could be part of proof of infringement, yes23:43
wolfspraulno really23:43
wolfspraulthat's why I say let's stay real23:43
wolfspraulSisvel is real23:43
wolfspraulthey are aggressive23:43
wolfspraulbut also professional23:43
wolfspraulthey will not use dirty tricks23:43
wolfspraulthat's my judgment23:43
wpwraktheir whole business model is one big dirty trick23:44
wolfspraulwell OK, fine, but you understand my point23:44
wolfspraulmaybe some mafia lawyers will feed data they gather from evidence to the hitmen.23:44
wolfspraulin my experience, Sisvel is by far not like that23:44
wolfspraulthey are using the legal system, which they did not create (although they spend a lot of resouces on lobbying to make sure it stays the way it is, and is modified in their advantage)23:45
wolfspraulso back to reality23:45
wolfspraulwiki/fora - no obvious infringements!23:45
wolfspraulno instructions on how to make mp3 play23:45
wolfspraulrafa: what wpwrak outlined above should work, imo23:46
wolfspraulthis is a decision to be made by tuxbrain anyway23:46
wolfspraulif mp3 capability is in nand, that's the #1 way to get caught23:46
wolfspraulwe learnt that already23:46
wolfspraulif it's installable - same, as long as they can proove that they stay within your system23:47
wolfspraulif there are instructions on website you operate - same23:47
wolfspraulthat's enough23:47
rafawolfspraul: what exactly? ("what wpwrak outlined above").. sorry, I am reading the whole buffer because I just came :)23:47
rafaso I was not following completely23:48
wpwrakand the same for technical workarounds. and trying to beat them at legal workarounds meet them at their home turf, so not a good idea either23:48
wolfspraulWerner is best at short summaries.23:48
wolfspraulno no, let's not try that23:48
wolfspraulthat's a separate battle23:48
wpwrakrafa: just the summary i posted after you came back23:48
rafawpwrak: ahh.. okey..23:49
wpwrakwolfspraul: btw, my questions for specific technical details are to define clear guidelines that can be used without requiring people even less familiar with all this stuff to make judgement calls for what may be very common tasks23:49
wpwrakrafa: lemme write this here :) the known danger areas are mp3, mpeg4, and h.26423:51
rafawolfspraul: yes, i know this.. but thanks a lot for discussing this. We have another good idea.. I think that it could be easy to have a twin image without links or software which can encode decode mp3/mp423:52
rafawolfspraul: wpwrak : we have that idea (a twin image) because you suggested that :) and I think that I could work on that easily23:52
wpwrakrafa: if patent issues pop up somewhere else, one would have to react to that. there's no way of telling. but then, sharism/tuxbrain/etc. wouldn't be a likely target for a new patent attack. they go after big companies.23:52
wpwrakrafa: (twin image) perfect. also, please keep in mind that the list of packet repositories the installer uses must not contain the "full" repository you'll probably have at jlime.com23:54
wolfspraultwin image?23:54
wolfspraulsounds suspicious :-)23:54
rafawpwrak: well, maybe we can provide an image without links to repositories at all.. So that would be really easy to get. At least users could test it.23:55
rafawolfspraul: twin image: I mean,23:55
wpwrakwolfspraul: the version to host at qi-hw :)23:55
rafawolfspraul: to provide a jlime image without links, without repositorires set, without software which can encode/decode mp3/mp4/patented codecs. This twin image would be useful to be hosted in other places with risk, like qi-hw servers ;)23:56
wpwrakrafa: naw, i think the best would be to build all the packages, have alternative packages for things like the package containing the list of sources, the rootfs, and anything that may incorporate the codecs, and copy this to qihw23:57
wpwrakrafa: if you really want to be pedantic, you could of course run two completely separate builds, even if you'd then also have a qi and a non-qi version of sed, vi, emacs, gcc, etc.23:57
wolfspraulthat's why openwrt introduced the PATENTED flag23:57
wolfspraulit's hard to single out packages if they are all tied together with dependencies23:58
wpwrakrafa: i think some of the key values of jlime is the large set of packages you can install, so they should be available on qi-hw too23:58
wolfspraulwpwrak: your idea about spreading patent knowledge is nice, but hard to execute23:59
wpwrakwolfspraul: that's an implementation detail :)23:59
wolfspraulbecause there are so few hard rules in the patent mess23:59
wolfspraulit depends on the enforcers23:59
wpwrakyup, unfortunately23:59
wolfspraulwe just know sisvel & mpeg la relatively well23:59
wolfspraulbut I'm sure there are other sisvels, especially in smaller fields23:59
--- Tue Oct 5 201000:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!