#qi-hardware IRC log for Wednesday, 2012-01-25

wolfspraulthe keyboard patches caused quite a stir on the list00:07
wolfspraulI cannot immediately respond to this now00:07
wolfspraulseems all sorts of conflicts there :-)00:07
wpwrakkeyboard patches ? looking ...00:08
wolfspraulsome people have no time, wish they had more, push feature-adoption to infinity, etc.00:08
wolfspraulI will just wait a little and uptick again - it's easier now00:08
wolfspraulthe KiCad core team first has to sort out some internal things00:08
wolfspraulwhat Wayne had responded with initially would have made any politician proud00:09
wolfspraulI think for now the patches will be moving nowhere00:09
wpwraknice reaction by Hoteev Sergey ;-)00:10
wpwrak"It is FUTURE!" :)00:10
wolfspraulyeah but there seem to be serious conflicts internally00:10
wolfspraulthat would also explain the inconsistent renames I'm seeing00:10
wpwrak(wayne) you mean the reply from last year ?00:11
wolfsprauland that separation vaguely described there will never happen, I don't even need to ask for specifics because it's obvious they just want to make the issue go away00:11
wolfspraulno, the new stuff00:11
wolfspraulsome people seem to have too much time/commitment to officially call it quits, yet too little to make solid changes00:12
wpwrakhmm, don't see it in the thread00:12
wolfsprauland they cannot decide which way to go ;-)00:12
wpwrakwhich subject ?00:12
wolfspraulJan 20, same subject00:12
wolfspraul"I apologize for not responding sooner" (not really needed when I needed a year to respond :-))00:13
wpwrakah, he broke the thread00:13
wolfspraul"hold off until we separate the underlying object code from the UI code and implement it as a DLL/SO"00:13
wolfspraulthat must be from the "how can I talk my dumb manager into leaving me alone hacking" seminar00:13
wolfspraullater in the thread he admits that that magic "separation" is at least 1 year out00:14
wolfspraul"at least" :-)00:14
wpwrak"It is at best a year out." ;-)00:14
wolfspraulI could restructure the patches in many ways, but there is no guidance/leadership at this moment.00:14
wolfspraulyet the magic 'separation' will still not fall from the sky...00:15
wolfspraulso I plan to do nothing right now00:15
wolfsprauluplevel once in a while00:15
wpwrakyeah, that may be the best approach00:16
wolfspraulit's a bit worrisome at least the way Wayne describes it that the "other lead developers" (all?) seem to all want to push this out a year or more?00:16
wpwrakthere's a lot of things they plan to change. e.g., the board file format. that's been pending for well over a year as well00:17
wolfspraulthat could make someone think KiCad development has haltet...00:17
wolfspraulyes but maybe all those things are actually in "x years" status00:17
wolfspraulmakes me proud of the Ben NanoNote :-)00:17
wpwraki guess if people get too restless, a fork may happen at some point in time00:17
wolfspraulit cannot survive a fork, because there is already geda00:17
wolfspraulthe footprint library is in abysmal state00:18
wpwrakso far, it hasn't, but there's been noises several times already00:18
wpwrakoh, i don't see a problem with a fork, as long as there are people who are motivated to keep it going00:18
wolfspraulI think the osmo-sdr guys switched from kicad to geda because of better scriptability somewhere, but I think it was inside the design (not sure).00:18
wolfspraulyeah, but this is too distracting for me00:18
wolfspraulso I can only uplevel, a bit more regularly than before00:19
wolfspraulI can't get into the kicad thing now00:19
wpwrak(distracting) we can just wait and see. if someone forks and doesn't fall over his or her own feet too quickly, we can join the new crowd00:19
wolfspraulwell yeah, but this needs an active developer base of multiple people really intensively hacking on00:20
wolfspraulat least00:20
wpwraki guess it needs people who can really sink 100% of their time into it00:21
wolfspraulisn't it funny how he describes the 'separation'?00:21
wolfsprauland that is after a talk with other lead devs?00:21
wolfspraulwhat separation?00:21
wpwrakfor dick, wayne, etc., it's basically a weekend project00:21
wolfspraula bunch of C++ classes directly through the dynamic library?00:21
wolfsprauland then - which ones?00:21
wolfspraulthe GUI classes?00:22
wolfspraulthe entire current code structure doesn't lean itself towards any kind of 'separation'00:22
wolfspraulit's just handwaving really00:22
wolfspraul"please go away and don't ask us details. and come back as late as possible"00:22
wolfspraulactually I think the current code structure is not bad, I would just leave it like that and gradually cleanup further00:23
wpwraki think they want to separate the data model better from the GUI. right now, it's all mixed together00:23
wolfspraulbut the separation as it is described there will never happen, 100% wishful thinking00:23
wpwrakof course, whether a cleaner separation really improves things in the end also remains to be seen00:24
wolfspraulsure, but they haven't gotten beyond the "we should really have this" level00:24
wolfspraulbecause that's a quite well known problem00:24
wpwraksometimes, a hundred ugly lines are better than five million beautifully structured ones00:24
wolfspraulbut there is nothing I can do, really. just uplevel.00:24
wolfspraulthe patches will stay outside.00:24
wpwrakyeah. we can try to piss them off a little. make them feel their control might slip if they don't integrate those features :)00:25
wpwrakwell, a few of the responses already went in this direction ;-)00:26
wolfspraulit's a design program, so the focus will always be GUI00:26
wpwrakyes and no. there is a deeper layer that's not GUI-centric00:26
wolfspraulif the focus is completely scripted, one could write an entirely separate engine that directly modifies the files00:26
wolfspraulok I'm just starting to think for them00:26
wpwrakyes. that's what we currently do. and i don't mind doing that.00:27
wolfspraulso if the main focus of Kicad is to be a (manual) design program, then naturally it's ok that a few command-line options are being 'inserted' into the otherwise integrated codes, as the patches are doing00:27
wolfspraulwe could cleanup that 'insertion' to a really nice level00:27
wpwrakthe files are not extremely pretty but i've worked with worse00:27
wolfspraullike I said, if the focus of the entire binary is 90% manual visual editing, and 10% at most some (few) extracted codepaths accessible via command line, then that is just fine00:27
wolfspraulthe cmdline gives you access to maybe 1% of what kicad can do in the GUI00:28
wpwrakyes, the problem of the command-line patches is that they try to reuse a lot of what kicad has. that's why there's a conflicht00:28
wolfspraulthat's not a problem, that's by design00:28
wpwrake.g., fped just ignores all the kicad code base00:28
wolfspraulthe command line options just auto-execute the GUI for you00:28
wolfsprauloh sure, that's a separate tool00:28
wpwrakyes, but it's that design property that creates the conflict for the command-line patches, while i can dodge that with fped.00:29
wolfspraulactually the patches are quite clean00:29
wolfspraulthere's a global, and some ugliness in some dialogs, but these things could all be cleaned up easily00:29
wolfspraulwithout waiting for a separation that will never happen00:29
wolfspraulbut like I said, I will do nothing00:30
wolfsprauljust uplevel00:30
wolfspraulnot my battle00:30
wpwraknaw, their problems seems to be that you talk to their old code, while they'd wish you to talk to their new code instead00:30
wolfspraulnew code?00:30
wpwrakalas, the lack of a time machine makes itself noticed once more ...00:30
wolfspraulah, ok00:30
wolfspraulthat separatoin is really nonsense00:30
wolfspraulit will not happen00:31
wpwrakit may not be nonsense00:31
wpwrakthat's a possibility00:31
wolfspraulwell then it's already there00:31
wolfspraulthey have it all nicely in dialog classes00:31
wolfspraulit's all fine00:31
wolfspraulit's a GUI app00:31
wolfspraulsure they could move non-gui stuff in one corner, and gui stuff in another corner - but that is already happening00:32
wolfspraulany practical insertion of a few cmdline switches would still go right through this00:32
wpwraksure. it's just mismatched perfectionism. 00:34
wpwrakthe argument would be perfectly valid if they'd be working on a massive redesign that's to be completed soon. but there's no "soon" in there.00:34
wpwrakmaybe they're just old and slow ;-) clone sebastien, run a global FPGA to EDA substitute on the clone, and then let him loose on kicad, and we'll have a fork that outperforms all our wishes in a few weeks ;-)00:36
wpwrakanyway, i agree with your conclusion that upleveling is the best we can do at the moment00:37
wpwrakjust outsitting it doesn't work, because changes are happening in the code base00:37
wpwrakbut the future perfect code isn't a realistic target either. i.e., you can't do anything today to make your patches work with that00:38
wpwrakit's also good to have the functionality around, so that people can get a taste of what it is like00:40
kristianpaulwow i see now why other people uses geda :)01:55
kristianpaulbetter i dint tried learn kicad yet :)01:55
kristianpaulgood that geda have its BOM :)01:57
wpwrakhmm ? what did you see ? and what BOM ?03:39
DocScrutinizerwpwrak: kicad broke ??I ?05:35
DocScrutinizerrather [A-Z]{1;2}I05:36
DocScrutinizerIE API|ABI|UI|GUI|...05:37
wpwrakwolfspraul: ah, kicad ... bzr release 3551 ? i get  eeschema/netlist_control.h:99: error: ISO C++ forbids declaration of 'wxNotebook' with no type10:50
wolfspraulyes, strange10:51
wolfspraulsomeone else reported that already10:51
wpwrakwas there a solution ?10:53
wolfspraulah, typical kicad11:01
wolfspraulbetween 3351 and 3378, they changed a massive amount of includes from #include "" to #include <>11:02
wolfspraulbefore it wasn't thought through or clean, and neither it is now11:02
wolfspraulbut someone must feel better with <> now11:02
wpwrakso i should use 3378 instead of 3351 ? or is this unrelated ?11:10
wolfspraulone moment11:11
qi-bot[commit] Wolfgang Spraul: upleveled to bzr3378 (master) http://qi-hw.com/p/eda-tools/e19224611:32
wolfspraulcan you try this against 3378?11:32
wolfspraulbut you may still run into some problem, it seems you see the same kind of issues as someone else reported to me already, which unfortunately I don't see on my system11:33
wpwrakah, and i had a typo in my checkout command. 3551 instead of 3351.11:49
wpwrakso 3351 would probably have been fine. trying 3378 now ...11:50
wpwrakinteresting changes in your patch :) +-SET(CMAKE_RELATIVE_PATH_TOP_BINARY "/home/user/kicad/kicad.3378")11:51
wpwrak286 out of 286 hunks FAILED -- rejects in file common/Makefile11:51
wpwraklet's go back to 3351 ...11:52
wpwrakah, typo was elsewhere. i did check out 3351, not try to check out 3551 11:58
wpwrakso i'll just wait for the corrected 3378 patch set :)11:58
wolfspraulsomething wrong, hmm12:00
wolfspraulone sec12:00
wpwrakno rush. i still have the working old installation :)12:04
qi-bot[commit] Wolfgang Spraul: again, this time with correct diff (master) http://qi-hw.com/p/eda-tools/626e11112:20
wolfspraultry that against 3378 (but like I said, I still do expect problems)12:20
wpwrakpatch set applies. already a good start :)12:26
wpwrakmake is running ...12:26
wpwrakbuild was successful !12:32
wpwrakinteresting. eeschema looks quite different now.12:33
wolfspraulok great12:33
wpwrakthe new logos look a bit odd ... but at least the BOM is now really easy to find ;-)))12:35
wpwrakthe rest seems mainly window dressing12:38
wpwrakbut great that it works. thanks !12:39
qi-bot[commit] Werner Almesberger: kicad-patches/README: updated for bzr 3378 (master) http://qi-hw.com/p/eda-tools/b8ab60a12:40
wolfsprauldo you think they are doing substantial development, aside from moving code around?12:55
wpwraktwo minor enhancement this year, it seems: the track width adjustment and the electra .ses files support, whatever they are13:08
whitequarksigh. damned flash. they introduced a lookupswitch opcode (a jump target lookup table), only to compute indices for it with this: http://pastie.org/324968313:16
whitequarkI am especially amused by the bottom of the construct.13:16
wpwrakwhitequark: if you take a job at the sewers, why do you complain about the smell of feces ? :)13:17
whitequarkoh, my job is to draw a map of pipes. employer never said anything about feces :/13:18
whitequarkthis is quite fun actually, and the code can and will be reused (most bytecode stack-based VMs are surprisingly similar)13:19
whitequarkhttp://pastie.org/3250093 Y U NO CONSTANT FOLDING?14:39
larscsometimes you really have to wonder14:47
whitequarkabout what?14:47
larscabout flash14:48
whitequarkconstant folding is ~150 lines in ruby, including opcode type inference14:49
whitequarkwhy the hell cannot they do that.14:49
larscdo they do any compile time optimizations at all?14:50
whitequarkAlchemy (LLVM AS3 backend) beats their AS3 compiler in, well, most of cases.14:50
whitequarkand it uses things like virtual stack and RAM.14:50
larscmaybe their vm does all the optimizations?14:52
wpwrakmaybe they simply don't care ? ;-)15:19
whitequarklarsc: the vm (tamarin) sucks quite a bit too15:23
kristianpaulwpwrak: about BOM i just read that from geda webpage15:42
wpwrakyou mean they only recently added BOM generation ?15:44
kristianpauldunno if recenlty just did a grep to its website15:45
kristianpaulcurl  www.gpleda.org/  | grep BOM15:45
kristianpaulno more :915:45
kristianpaulactually i tought you already knew it about it 15:46
wpwraki'm not tracking geda15:52
kristianpaulinteresting http://geda.seul.org/wiki/geda:bom_readmeok15:54
kristianpaulok 15:54
wpwrakthanks !15:54
wpwrak"Eventually I'd like to integrate this with some sort of a database" hehe :)15:55
wpwrakwell, every journey of a thousand miles begins with the first step ...15:55
tediousHi, been looking around the wiki, and web searched a bit, but couldn't find the answer. Forgive me if it's obvious or easily found. - Does the Ben NanoNote require any non-free(non-libre) software?19:21
virictedious: no19:42
virictedious: I wrote a small distro for it - zero blobs, as far as I understand19:43
viricthe sound driver is somehow 'big', but it's in the linux kernel tree, so I imagine it's free.19:43
tediousviric: Thank you for the response, this is great to know. - May I get the sound drivers name?19:44
viricjz47xx is the SoC19:44
viricthen, audio for that19:44
virictedious: the kernel options are: SND_JZ4740_SOC and SND_JZ4740_SOC_QI_LB6019:45
viricnow that I check, they are not very big19:47
tediousHmm, being a newbie, not entirely sure how to ask this, and if I ask it incorrectly, I do apologize. - How would I check the license for the drivers? :o19:54
virictedious: read the acompaning licence in the header of C files of linux sound/soc/jz474020:03
viricI'd do that.20:04
tediousviric: Thank you. :-)20:54
tediousAs for the driver needed for the video portion, what would it be called?21:34
viricthe frame buffer21:35
viricwhat is this investigation about? :)21:35
whitequarklook for jz4740_fb.c21:35
tediousviric: I want to be _certain_ that the next hardware I purchase doesn't require non-free software. :-)21:37
tediouswhitequark: Thank you.21:37
viricI built the distro I told you from source.21:38
viricI couldn't see any magic in the path21:38
kristianpaulviric:can i boot your marvelous distro by micro-sd?21:39
viricit will give you a mingetty and then a bash :)21:40
viricwhat does it mean, by microsd?21:40
viricwill it load the kernel from the microsd?21:40
kristianpaulmemory card21:40
tediousviric: Not that I didn't believe you, but I wanted to seek it myself as well. :3 - Also was curious about the licenses. - What is your distro name, by the way?21:40
kristianpaulno the distro i mean, full userland21:40
virictedious: world famous nanonixos21:40
virickristianpaul: I don't think it will boot with an openwrt kernel21:41
viricI use /init as boot.21:41
viricinstead of /new-path-thought-by-openwrt21:41
virickristianpaul: that, imagine said "con rintintín"21:41
kristianpaulje ;)21:45
viricah, it loads the kernel from the sd21:46
virickristianpaul: so I could try to prepare a tarball to unpack into a sd, rigth?21:48
viricwell, there isn't much to see, really... in the best case, it will boot and offer you a bash.21:50
viricyou'll type lynx, and you'll see lynx21:50
virickristianpaul: the nanonote web page has a section "Is this for me?" :21:51
tediousSo, is there any hope for an external keyboard on the ben nanonote? (Been looking around online and not getting much to go off of)21:54
virictedious: if you could find a sdio keyboard.. :)21:55
xakhhey, just a small question21:55
xakhI keep hearing rumours that the Ya is like, in production.21:55
xakhis this true?21:55
xakhwolfspraul: you know anything about the timeline of Ya nanonote?21:57
xakhzear: you know anything about the Ya Nanonote?22:06
wolfspraulwow we have rumors like at Apple - cool22:09
wolfspraulwhat do we do to make the rumors get bigger? we say nothing22:10
wolfsprauloh wait, we are not Apple22:10
wolfspraulxakh: the rumor is wrong. I think the Ya will not happen in 2012, at least my full focus is on Milkymist One now, whose technology may well be merged with the NanoNote (then Ya) later.22:10
wolfsprauluntil then it's all about Ben NanoNote, which is very much alive, shipping, software being improved, etc.22:11
wolfspraulthanks for asking!22:12
xakhI have mine, and my friend was eying one22:12
xakhalright, thanks for clearing that up!22:13
kristianpauli want it ya ;)22:13
wpwrakya ya ;-)22:19
tediousviric: Having no luck finding an sdio keyboard. :p22:43
tediousKnow of anyone making/using one successfully?22:44
mstevensI should dig my nanonote out again and play with it22:45
mstevensupdate the build22:45
viricno, I don't know any22:49
viricbut I never searched22:49
viricmstevens: all fine with the fuloong?22:49
mstevensviric: various minor problems22:50
viricfixed problems?22:51
mstevensnot yet22:51
viricI finally booted the kernel.... enabling VGA_CONSOLE hangs it early at boo22:51
mstevensviric: worst problem is extremely poor DVI video quality22:51
viricI've never had a dvi cable23:05
viricthe vga has some problems at very high resolutions23:05
viricsome trembling23:05
viricisn't dvi a digital link? What can be bad there?23:06
viricmaybe you use the analog signals of the DVI?23:06
mstevensI'm seeing weird stuff23:06
mstevensincorrect colours23:06
mstevensmoving texture in what should be solid colour blocks23:07
viricand the vga is fine?23:07
virichow weird then. maybe noone used the dvi.23:07
mstevensviric: I'm wondering if it's software, I'm wishing I'd tested more carefully with openrays or whatever it is before I wiped it23:11
viricwrite to the list23:12
mstevensit's hard to see what it could be though23:12
mstevensviric: I may try that23:12
--- Thu Jan 26 201200:00

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