#qi-hardware IRC log for Friday, 2010-10-08

wpwrak#2 enumerates, too. yeay ! :)00:02
wpwrak(component file) no, you can move pins00:02
wpwrakclick on the circle00:03
wpwrakeor just move there and type M00:03
wpwraksome fine day, when qi-hw gets bought by some ultra-conservative megacorp, and they find the USB IDs, all nicely written in "leet", that will drive them crazy ;-)01:33
tuxbrainhey hey wait, I have just read it, Fedora electronic lab and Milkymist, what it means a)FEL will run in milkymist or b) You can work on milkymist design with FEL? whatever option is cool indeed01:55
wpwrakshould be b)01:56
tuxbrainwell I think is time to think about a FEL post in tuxbrain :)01:57
tuxbrainmaybe become a fedora Ambassador02:07
qi-bot[commit] Werner Almesberger: Increased clearance from 9 mil to 9.7 mil. http://qi-hw.com/p/f32xbase/ad0015d02:56
qi-bot[commit] Werner Almesberger: Make the board a bit bigger for more relaxed spacing of traces and components. http://qi-hw.com/p/f32xbase/c7a53e802:56
qi-bot[commit] Werner Almesberger: R1/R2 were very close to U1 - move them away. Some small cleanup. http://qi-hw.com/p/f32xbase/807616f02:56
qi-bot[commit] Werner Almesberger: hw/cam/: toolpath generation for cutting the PCB. http://qi-hw.com/p/f32xbase/c9f4c9b02:56
qi-bot[commit] Werner Almesberger: Customize the FT232 EEPROM with the Qi Hardware ID and other parameters. http://qi-hw.com/p/f32xbase/9777d8802:56
qi-bot[commit] Werner Almesberger: Added BOOKSELF and dsv setup procedure. http://qi-hw.com/p/f32xbase/3fd172002:56
djbclarkwolfspraul: ah syncronicity08:15
wolfspraulhow are you doing?08:16
wolfspraula little worried about you :-)08:16
wolfspraulwe are moving along the lines we discussed back in July08:16
wolfspraulbtw that guy Henry went dark on me08:16
djbclarkwolfspraul: I am, in fact, still alive. Finding a fulfillment house willing to work in low quanities in US near me has so far been fail. Just brought over 50 Ben Nanonotes to contractor last week. Should be selling later this month.08:16
wolfspraulwell, let's say he exactly met the expectations I had into his "I will release code next week"08:17
wolfspraulI'm around the block for a while... :-)08:17
wolfspraulmeanwhile we are actually moving, and code is free & open from day 1, how it should be08:17
wolfsprauldjbclark: perfect, good to hear that08:17
wolfspraulI'm in for the long run, so no worries08:17
djbclarkwolfspraul: Hallam? He is also still alive. I wasn't aware he had promised anything.08:18
wolfspraulwe are trying to spend a bit more time to write up good readable news, with pictures and all08:18
wolfspraulthat will also improve the communication among the core members like you08:18
wolfspraulotherwise our project is too diverse08:18
wolfsprauldjbclark: sure we quipped over it back then.08:18
wolfspraulit's not his fault08:18
wolfspraulhis employer won't let him release nada zip anything08:18
wolfspraulthat simple08:18
wpwrakwhat wonderful code shall we expect in 4 months ? (standard development time adjustment: add one, duplicate, then convert to the next higher unit)08:19
wolfspraulGPS firmware08:19
wolfspraulthe stuff kristian paul works on08:19
wpwrakooh ! that would be nice08:19
wolfspraulsanta claus also could remember the copyleft iphone this season...08:20
wolfsprauldjbclark: if you are using an RSS reader, please add http://en.qi-hardware.com/feed/rss20.xml08:20
wolfsprauland help spreading the link08:20
wpwrakis kristianpaul the guy who's doing GPS in an FPGA ? the project where we looked at the RF frontend chips ? i thought that was someone else08:20
djbclarkwolfspraul: ah yeah I thought the useful stuff from him was the hardware component descrptions maybe; the code sounded like it would at least be in purgatory (where there are, of course, lots of attorneys) for months at minimum08:20
Action: djbclark -> breakfast food08:21
wolfspraulwpwrak: yes, http://en.qi-hardware.com/wiki/GPS_Open_Stack08:21
djbclarkwolfspraul: if by "using" you mean "have", sure, if you mean "actively have read anythin in the last few months", less so :-/08:21
wolfspraulhe, OK08:22
wolfspraulstill spread that link if you can...08:22
djbclarkwolfspraul: sure, I'll go dent it...08:22
wolfspraulyes! :-)08:23
wolfspraulthank you!08:23
wolfspraulthat's the pure original qi news link08:23
wolfspraulno aggregated stuff08:23
wolfspraul100% copyleft hardware :-)08:23
wpwrakwolfspraul: ah, nice. thanks ! i've been wondering if anything had come out of it. good to see that it's alive :)08:24
wolfspraulhuh of course08:24
wolfspraulyou are chatting with him every day :-)08:24
wolfspraulwpwrak: you are Werner, right?08:24
wolfspraulit's the same guy working on the case for ben-wpan with his reprap08:24
wolfspraulhe got a sige 4162 evb, and an antenna, and some other smaller stuff08:25
wolfsprauloh and a SIE board08:25
wpwrakwolfspraul: hehe ;-) i just didn't know he was the one doing that GPS thing. he occasionally mentioned something about GPS, but relatively little and nothing that would "connect" it.08:26
wolfsprauloh, I see08:26
wpwrakworld view updated :)08:27
qi-bot[commit] Werner Almesberger: fw/c2usb.conf: USB product ID assignment is now official http://qi-hw.com/p/f32xbase/77d684808:29
wpwrakgrr. and i need caffeine. s/fw/hw/ # grmbl08:29
kristianpaulwpwrak: yes10:25
kristianpauli just dont like spread loudly what is not done yet :)10:25
kristianpaulalso is a huge learning curve, task i still doing almot every day10:27
kristianpaulright now i want debug the GPC IC frontend output before go further10:27
kristianpaulthats why i want use SIE as scopemeter10:28
kristianpauland is not GPS on FPGA initally (surelly will, but who knows)10:28
kristianpaullets see how the ben can handle the GPS correlatiion/tracking  and no die in the try ;)10:29
wpwrakkristianpaul: will you generate/collect a sample data stream to try on the Ben ?10:38
wolfspraulnice idea10:38
wpwrakkristianpaul: (scopemeter) hmm, but you have a digital output on the RF frontend. so why the scopemeter ? (besides generally being a useful thing to have around)10:38
wpwrakwolfspraul: much nicer if the data you're trying to reproduce that bug with doesn't change every second ;-)10:39
wolfspraulwell I also thought about it from a practical standpoint10:40
wolfspraulright now we only have one sige 4162 evb10:40
wolfsprauland they are hard to come by10:40
wolfspraulbut if kristianpaul uploads sample data streams, more people can join development10:40
wolfspraulwe 'virtualize' the evb :-)10:40
wolfspraulit also reminds me of a meeting I had at u-blox a while back, where they told me how some cameras had a system where the camera only stored the raw data stream when taking the photo, and it would be converted to the GPS location only later when the pictures are copied to the desktop10:41
wolfspraul(of course all with proprietary software)10:41
wolfspraulthat way the cost of the gps solution in the camera would be much lower10:42
bartbesbut crappier10:43
bartbesas you would be tied to their software10:43
wolfspraulsure, although that is not the fault of the hardware10:43
bartbesI know, it is simply a note regarding that particular product10:43
wpwrakwolfspraul: (board access) yes, good point10:43
wolfspraulon the camera itself there are indeed few applications where the user would want to know the gps coordinates immediately10:43
wpwrakbartbes: free the software and it's uncrappy again :)10:44
wolfspraulthat's something they noticed, so the camera is simplified (the hardware), without the user noticing a lost feature10:44
wolfspraulthat in itself is a good realization10:44
bartbeswolfspraul: well, if I have a camera I want to be able to take out my mem card and copy the photos, in this case with the gps position in it10:44
wpwrakwolfspraul: and since we're all about making things chin^H^Heap .. :)10:45
bartbesthough if the gps data is stored on the mem card it is already much better10:45
wolfspraulI don't know how they embed the data with the picture, in this case the raw data instead of the gps coordinate10:45
wolfspraulanyway we are diverting. it just came to my mind when werner spoke of 'sample data streams'10:45
wpwrakbartbes: is there actually a "standard" way to put GPS data into EXIF ?10:45
bartbesI believe there is10:46
wolfspraulbecause essentially a 200kb data stream (a vague memory from that meeting) is the same as a gps coordinate10:46
wolfsprauljust needs a bit of math :-)10:46
wolfspraulso where you move that math depends on the application/device type - why not...10:46
wpwraksmall details :)10:46
bartbeswpwrak: though if I remember correctly there are a few different formats, and some of them are actiually stored in the EXIF comment section..10:46
wpwrakwolfspraul: also helps a lot if you can debug your math on the PC10:46
wolfspraulbartbes: I am not saying I like the pile of proprietary crapware that will stand at the end of this, we are on the same page about that.10:47
wpwrakbartbes: good. so at least some made a reasonable choice :)10:47
bartbeswpwrak: though wikipedia seems to *suggest* there is a standard way of adding it to ExIF10:48
wpwrakwolfspraul: (proprietary) it's at least a choice that can be useful to have. besides, coming from us, it would be open anyway. so worst case, you'd need another open utility to decode the samples10:48
wolfspraulsure, like I said it's not the fault of the hardware10:49
wolfspraulthe idea to drastically simplify the gps module/hardware in a camera because in that device it's enough to collect the raw data stream is a valid idea, I think10:49
wpwrakwolfspraul: definitely. just like SDR moves the hard stuff into the CPU. so you move everything one further step down, towards less specialized equipment.10:52
aisahaving worked with OpenWrt to build 1.5 packages, I have to say this system is really great.11:08
wolfspraulaisa: did you have a chance to look into that owners/permission problem, whether you still have it?11:14
wpwrakwolfspraul: btw, how about the ghostbusting band-aid ?11:16
aisawolfspraul: porting this package, jbofihe, has taken longer than I planned.11:17
aisathe program compiles part of itself and then uses that to finish the build,11:17
wolfspraulwpwrak: another mail in my inbox11:17
wolfspraulit's excellent. if it works we're done.11:17
aisaso I've had to look at gforth package and do a lot of searching to get it working.11:17
aisaI am planning on testing this permission problem next, after this package is successfully ported.11:17
aisaI'm happy to report that as of this morning it is running on the NanoNote,11:18
aisabut I still have at least one bug to fix.11:18
wolfspraultoday was not a good day, I ran out of battery without noticing, after power cycle some files were corrupted (executables), I thought a few quick apt-get would fix it, but maybe I should not have been so quick as the problem got bigger, and so on11:18
wpwrakwolfspraul: was that about the band-aid ? if yes, i'll switch the heap randomization off and rebuild avt2. then we'll see the truth, ugly or not11:18
wolfspraulof course11:18
wpwrak(fs corruption) eek11:18
wolfspraulthat's what I would do once I get to your mail11:18
wolfspraulamazing that you (think you) tracked it down so quickly. if that workaround works we're done.11:19
wpwraki still don't know what really causes it. but for that, i'd also have to compare kicad versions and so on. but that can wait :)11:20
wpwrak(quickly) not really. albout a week :-( well, with several interruptions.11:21
wolfspraulwe document it properly, and that's it. after a year we can check whether it's still needed.11:21
wpwrakhehe :)11:21
wpwrakrebuild is running. let's keep our fingers crossed :)11:22
wolfspraulmaybe uninitialized variable somewhere? small compiler differences in the myriad of C++ constructor options11:26
wolfspraulthere are many ways something like this can creep in11:26
wolfspraulthe workaround (if it's stable and works) is fantastic I think11:26
wpwrakyes, my guess would be some uninitialized boolean for boldface on/off11:26
wolfspraulwpwrak: do you know the 1000 different ways of C++ constructors?11:27
wpwrakwolfspraul: i'm not sure i want to ;-)11:27
wolfspraulthat's a real jungle. I do not envy a compiler writer who has to implement this bug-free, somehow.11:27
wolfspraulyou could easily study just constructors for a year or two11:27
wpwraki didn't know there were such a mess. thought they just get called and that's it11:28
wolfspraulamazing how a simple concept such a linear memory can be made so complex11:28
wolfsprauloh no11:28
wolfspraulbut let's not continue the C++ rant :-)11:29
wolfspraulif that workaround works it's great!11:29
wpwrakit's done. how do we like this ? http://projects.qi-hardware.com/schhist/board-qi-avt2/11:31
kristianpaulwpwrak: (scopmeter) sure the SiGE EVB have pulse digital ouput, i'm just want debug the signal11:35
wolfspraulperfect I think!11:35
wolfspraulwpwrak: I'll document this tomorrow, thanks a lot for the help!11:36
kristianpaulwpwrak: in teory this si square ouput, but it could get triangled due some reason related to capacitance load (i need read more about it)11:36
kristianpaulwolfspraul: yup thats the plan, upload the streams11:37
kristianpaulsurelly :)11:37
wpwrakwolfspraul: two more steps: 1) rebuild the rest too (there may be ghosts we haven't spotted yet), and 2) set things such that this survives a reboot11:37
kristianpauli just want to make sure have a clean data aqusition process usign the Ben11:37
wolfspraulwpwrak: I'll do that, no worries11:37
wpwrakkristianpaul: ah, just the signal integrity. i see. hmm, what sample rate do you expect to get ?11:38
wpwrakwolfspraul: ok, so i'll leave all that to you :) thanks11:38
wpwrakkristianpaul: (on the scopemeter)11:38
kristianpaulwpwrak: (generate data) yes is other idea simulate QI data froma fpga to the ben in order to test benchmarking11:39
kristianpaulwpwrak: (rate) there are 3 signal out, SYNC(2Mhz) , DATA (4bits per sync), CLK (8Mhz)11:41
wpwrakyes, that too. ah, you'll still need to make that spi-to-spi cpld :)11:41
kristianpaulbut i want see the sqare behavior before11:41
kristianpaulthats why i need the scopemeter11:41
kristianpauli plan ignore the CLK11:42
wpwrakkristianpaul: (rate) for signal integrty you need some multiple of that. some recommend 10x, but that's probably a bit excessive. still, 80 MHz is achievable.11:42
kristianpauland just track SYNC and do internal timingns to get the right 4 bit data11:42
kristianpaul4 bit pulse data**11:42
kristianpaulwpwrak: you mean i need 80Mhz scopmeter to measure 8Mhz data11:43
wpwrakwhy not get a decent bench scope ? :) there's an advantage in having a device with real controls that doesn't depend on the pc ...11:43
kristianpauli dont have all that money11:43
wpwrakyeah. well, it actually depends on the raise/fall time specification.11:43
wpwrakisn't this a for-work project ? :)11:44
wpwrakone that's part of your job11:44
wpwrak(dunno what you're doing, actually)11:44
wpwrakah, pity11:44
kristianpaulmy real-work is information security11:45
wpwraknot much hardware then11:45
kristianpaulthats from all the money came from to bvuy reprap, fpga abnd otehr stuff11:45
wpwrakbut you can cupcake keys to the server cabinets ;-))11:45
wpwrakyou should ask for a raise. explain them that the expensive hobbies make you more qualified ;-)11:46
kristianpauli considered thata ctually11:47
kristianpaulwpwrak: any way i like support this project :)11:53
kristianpaulwpwrak: in case my DIY scopmeter fail i think i could byt this http://shop.ebay.com/i.html?_nkw=scopemeter&_sacat=0&_odkw=scope&_osacat=0&_trksid=p3286.c0.m270.l1313#11:53
wpwrak100 MHz, good price for that11:55
wpwrakbut i wonder if the price is for real. that thing costs USD 2600 new.11:55
wpwrakso if they sell it that cheaply, they either have no clue or it's broken11:56
wpwrakah, it's an auction. that's why.11:56
wpwrakamazing how the prices differ. one place sells it at USD 2100 refurbished. another has it at USD 1500, apparently new.11:59
kristianpaulwait where is 100Mhz i read it was 10Mhz..11:59
wpwrak10 MHz would be completely useless :)12:00
kristianpaulit is12:00
kristianpauli need go to a lab then12:00
wpwrak10 MHz would mean that you can see a 10 MHz sinus. so forget about the signal integrity of anything faster than 2 MHz :)12:01
qi-bot[commit] Juan64Bits: Routing image sensor. http://qi-hw.com/p/xue/ad23f9e12:43
qi-bot[commit] Juan64Bits: Cleaning. http://qi-hw.com/p/xue/cc01f3c12:47
kristianpaul2000usd !!13:39
kristianpauli think i need ask for a electronic lab in the local university13:39
qbjectaisa: does your work on wordgrinder mean that it can now be installed on a nanonote?13:50
aisait can.13:55
aisaI should have marked the package as unbroken.13:55
aisait did not work for me on the factory image,13:55
aisabut once I flashed to the latest build.13:55
aisawordgrinder works.13:55
aisaI've not really exercised it, but it opens and writes.13:55
aisaqbject: that last bit was all for you.13:56
qbjectCool. Thanks.13:58
qbjectaisa: I realized after I asked that I should probably avoid using it and keep to my own project, but it's good to know that it's out there. :)13:59
aisahere, let me port more software for you ;-)13:59
qbjectHeh. Well, now that you mention in... :-P14:00
qbjectin -> it14:00
wpwrakkristianpaul: do you have it far to bogota ?14:01
Action: aisa is watching http://www.youtube.com/watch?v=9u-MczVpkUA14:11
aisaafter a post to the mailing list.14:11
qbjectaisa: how's it looking?14:56
aisaqbject: what, the video?14:58
aisaI cross posted it to another list, I enjoyed it.14:58
qbjectVery cool. I'll watch it tonight while I'm duct taping Dingoo libs into my nn.15:00
aisaI spent a lot of time getting a package to generate files on my host, only to realize they were platform dependent and didn't work on my NN.15:01
aisaI did get a package out of it, which I can use to generate those data files.15:01
aisathat's my next duct tape project.15:02
aisaI know nothing about Dingoo...15:02
Action: aisa goes to investigate.15:02
qbjectaisa: NN's less productive but no-less motivating bretheren.15:03
aisaah, cool.15:04
aisaDon't get too caught up playing POWDER ;-)15:04
qbjectNever tried it, but okay.15:05
aisait's a roguelike for platforms like the dingoo.15:05
qbjectAch. Meritous is the crawl I'm pursuing atm.15:06
qbjectAnd OpenTyrian.15:06
qbjectAnd other various modern interpreters for old content.15:07
qbjectaisa: oo. POWDER does have the sort of pixels I love.15:15
aisaafter reading about Meritous and OpenTyrian I had to forget they existed.  At least for now ;-)15:16
aisaI do plan on porting POWDER to the NN.15:16
aisaNot sure of my timeline yet.15:16
aisait's not in my current sprint.15:17
qbjectaisa - before you get carried away, I think zear has already done so, though I don't know the status of his port: http://downloads.qi-hardware.com/people/zear/games/15:20
qbjectIf you REALLY want to have to forget that something exists, take a look at ASCIIpOrtal15:20
Action: qbject downloads15:21
kristianpaulwpwrak: 12 hrs by ground16:11
kristianpaulwpwrak: i live in Buga, is like 450~ away16:12
kristianpaulwpwrak: why?16:12
kristianpaulaway from here*16:13
kristianpaulwpwrak: if i dont get the ost of scopemeter i'll trust signaling (glup!) and move to a logic analizer or sort of..16:34
kristianpaul09:42 < wolfspraul> it also reminds me of a meeting I had at u-blox a while back, where they told me how some cameras had a system where the camera only stored the raw data  stream when taking the photo, and it would be converted to the GPS location only later when the pictures are copied to the desktop16:38
kristianpaulyeah thats reelly usefull16:38
kristianpaulas cameras can correlate all that data :)16:38
kristianpauli'm not sure but u-box seems to use the same SiGE chip i have or same family**16:39
kristianpaulwpwrak: The CLK_OUT, DATA and SYNC output drive is set to maximum by R6 = 0R pulling SE4162T RVI pin 1216:45
kristianpaulto LDO_OUT pin 24. RVI pin can be left open circuit for minimum output drive (7.5pF load). The GPS16:45
kristianpaulclock and data outputs are slew rate limited to minimize coherent feed-round interference which would16:45
kristianpaulotherwise de-sensitize the radio. As the load capacitance increases the higher frequency clock will tend16:45
kristianpaultowards a triangular waveform. In most practical applications the output drive should be more than16:45
kristianpauladequate to drive the combined PCB and processor input load capacitance. However, problems may16:45
kristianpauloccur with large development platforms. In such cases a local, triple buffer for the clock and data lines16:45
kristianpaulmay be required. Consideration of the total load capacitance should be made when viewing the condition16:45
kristianpaulof the clock waveform with a standard 11pF scope probe.16:45
kristianpaulthat really worry me, thats why i need xscope, but i can guess this will happen and just implement they buffer they said..16:46
kristianpaulor avoid clock16:46
kristianpaulmay be16:46
kyaki just compiled a "Hello, world!" on Ben with gcc, under OpenWRT :)16:52
kyaknow to create a usable Makefile16:53
qbjectkyak: awesome.16:54
kristianpaulkyak: ohh?16:56
kristianpaulkyak:  ported gcc? how you achieved that?16:56
kristianpaulkyak: great :)16:56
lekernelkristianpaul, basically fighting with autocrap build system I guess :o)17:01
kristianpaullekernel: ;)17:02
wpwrakkristianpaul: (bogota) ah, i was thinking that carlos' group could help you with measurements. but 12 h is a little far.17:04
kristianpaulwpwrak: well i'm just finished writing a mail to a profesor i know from the local public univesity here,17:05
kristianpauli'm now they hav lot of scopemeters17:05
kristianpauli used soem of then17:05
kristianpaulbut i'm not stduying there17:05
kristianpaulso i asked him for some help and support17:06
wpwraklet's hope they can do something for you17:07
kristianpaul wpwrak if were in bogota (NOO ;)) i think anyway carlos keep busy with its courses and students so let him work :)17:08
kristianpaulme too :=17:08
wpwrakseems that you could just start with R6 = 0R. given that the data clock is constant, you'd notice quickly if anything is amiss17:08
wpwrakkristianpaul: carlos would not have to assist you _personally_ ;-)17:09
kristianpaulwpwrak: sure but i know equipment is expensive and he dont know nothing about me so.. at least that count17:10
kristianpaulanyway i'm not near to bogota to worry about it :)17:10
kristianpaulin the other hand this professor know me since years go so is different ask for soemthing like this17:12
wpwrakkristianpaul: for measuring the clock frequency, you can use the approach i used for ben-wpan/cntr: implement a large (> 32 bit) free-running counter, read it from a pc, and synchronize with NTP17:14
wpwrakNTP isn't extremely accurate bit it's immune to drift. so you can implement an arbitrarily precise frequency counter if you're willing to wait long enough17:15
kristianpauli'm doing !17:15
kristianpaulfor mlkymist17:15
kristianpaulbut is kind of crude yet17:15
kristianpaulbut can be improved :)17:16
wpwrak(already) great !17:16
kristianpaulhmm you gave me an interesting idea17:16
wpwraki used "cntr" to check that the transceiver has a clock that's better than +/- 30 ppm. i can get a measurement accuracy of a few ppm quite easily. a frequency counter with the same precision would cost something like USD 2-3k17:17
wpwrakof course, it would be faster :) 1 s vs. 20 minutes :)17:18
wpwraki wish my scope had a larger sample memory. debugging usb-driven communication (ft232 bit-banging) is quite da pain. 1 MHz output change rate with delays of several ms between bursts. i either see only a small fraction of even the simplest session or it's with an unusably low sample rate :-(17:28
wpwraksigh. why or why do those nice new rigols with 140 MSa have to be so expensive ...17:29
Action: kristianpaul sigh17:33
Action: kristianpaul checks bitscope.com17:43
wpwrakthat ft232 is a weird beast. seems that set_bitmode can bypass a write.18:53
wpwrakalso nice: in peak detect mode, my scope doesn't combine min/max samples but shows them one after the other, "inventing" nice glitches. haven't used peak detect mode for a while, but i think with older firmware, it didn't do this.19:37
wolfspraulwpwrak: the milkymist 'memory instability' bug is fixed, turned out to be a software problem in usb event handling20:54
wolfsprauljust fyi20:54
wolfspraullots of time went into this, but at least we got some stability improvements in the rc2 board out of it, that (assuming there are no regressions) should be a good thing in general20:54
wpwrakwow. sounds like a pretty non-obvious connection20:55
wpwrakyeah, sometimes you find a lot of other problems when chasing a bug long enough ;-)20:55
wpwrakwell, congratulations to the mikymisters ! :-)20:56
wpwrakwolfspraul: seems that you have enough stuff already for a community news update :-)20:59
wolfspraulit's a lot of work to turn it into high quality news though21:01
wolfspraulI think I will work on a milkymist press release for 27c321:01
wolfspraultry to get a quote from Lattice, etc.21:01
wolfspraulthat's another thing that was missing in the Nanowar release - a quote21:02
wpwrakmaybe a good strategy for the community news would be to have a frequency or quantity trigger. if you have enough items, don't wait for more. that way, you avoid getting a small encyclopedia every once in a while (which many people won't read with the same amount of attention something smaller would get).21:08
wolfspraultotally agree21:09
wpwrakthe frequency trigger would be a timeout to show that the project is still alive21:09
wolfspraulI may need to move some of the 11-1 stuff into 12-121:09
wolfspraulbut one by one21:09
wolfspraulalso, even though schhist, or 2.6.36 inclusion, would justify a proper press release of their own, I don't have enough time right now to do it21:09
wolfspraulso it will just show up as an entry in the community news...21:10
wpwrakcommunity news are fine for such things. they're not big enough to make it into, say, heise newsticker anyway21:12
wolfspraulI think we should only do a press release when we reach some minimum quality level, with regards to the text, pictures, quote, etc.21:12
wolfsprauland also on the release day of the press release we need someone full-time to follow up with media21:13
wpwrakyes, that too. and have something substantial enough. a product launch, for example. or some major new use(r)21:13
wolfspraulso unless we have that capacity, it's better to bring it nicely in the community news rather than have a lot of failed press releases21:13
wpwrakyeah, i saw how you explained it to tuxbrain. quite a bit of work :)21:13
wolfspraulbut think about how much work whatever the press release is talking about is21:14
wolfspraulif you want your press release to be picked up, you are competing with others and they put substantial efforts on getting their word out...21:14
wpwraksure. still, didn't expect that you'd actually need a propagation-around-the-world schedule. well, never thought about it.21:14
wolfspraulwell that's my suggestion and I have seen it like that in recent years. if someone knows a better way, please late us know...21:17
wolfspraulthese things are changing, it depends on how the media is working21:17
wolfspraulI remember even 10 years ago it was all about the address list (snail mail) you had21:18
wolfspraulso on the day of the press release, you would mail out x00 letters to the 'secret list' that only your PR guy had and would never share with you21:18
wolfspraulso at least that has been blown wide open now21:19
wpwrakit does sound quite reasonable. i'm more the "fire and forget" type. but i can see how that approach would fail miserably in this case.21:19
qi-bot[commit] Xiangfu Liu: add the manual page manual/fped.1 http://qi-hw.com/p/fped/f8ed57421:19
qi-bot[commit] Xiangfu Liu: add debian/fped.manpages  for install manpage http://qi-hw.com/p/fped/31b5a9d21:19
qi-bot[commit] Xiangfu Liu: add DESTDIR to Makefile. http://qi-hw.com/p/fped/d50bb0b21:19
wolfspraulwith a little time, you can easily collect submit and personal email addresses of any media you like21:19
wpwrakopen hardware, open PR ;-)21:19
wolfspraulI think something new in recent years is that stories are 'pulled'21:20
wolfsprauljournalists are so overloaded with bad quality stuff, that they rather contact known PR agents themselves, and work with them to craft/time interesting stories21:20
wpwrakheh :) interesting21:20
wolfspraulbut who knows, we just do step by step, see what works, adapt, etc.21:20
wpwrakand nice for the PR agents. strengthtens their position21:21
wolfspraulthey found a replacement for the 'address list' they guarded like a state secret 10+ years ago21:21
wolfsprauleverybody has that now21:21
wolfspraulbut the result is flood, and... things turn around and the PR pro is back in business... :-)21:21
wolfspraulback then they often were an overpriced mailing service, leaning back and sitting on their list21:23
wolfspraulI do think now they have to work harder on quality, so someone will contact them to pull stuff21:23
wpwrakfat cat biz :)21:23
wolfspraulif I had more money I would work with a PR agent right away21:24
wolfspraulworks like a multiplier on whatever you have21:24
wpwrakcould we have someone in the community who'd have channels ? there are some people who seem to have more of a journalistic profile than a developer profile21:25
wpwrakor perhaps you already have better access anyway21:26
wolfspraulhelp is always good21:26
wolfspraulI don't know, it's a good question21:26
wolfspraulxiangfu continues to work on the fped Debian package21:26
wolfspraulshould be almost done21:26
wolfspraulas for getting it into Debian, we might need a sponsor or so (I think first xiangfu uploads it to mentors.debian.net, he can do that himself, and after that it needs to be pulled up by higher authorities)21:27
wolfspraulI was thinking about contacting Jan Luebbe, he's a Debian developer and may have the powers21:27
wpwrakah, interesting process :)21:28
wolfspraulBas Wijnen is writing from a Debian address too, I don't know whether that means he is a Debian developer and wants to/can help21:28
wolfsprauljust fyi21:28
wolfspraulXiangfu knows someone too, who helped get xburst-tools to the next stage21:29
wolfspraulbut that took about 6 months I think, so this time I will try to find others as well21:29
wpwrakurgh. that's slow.21:29
wpwrakwell, debian has that reputation to uphold ... :)21:29
wolfsprauleventually we'll get it in, no worries21:30
wolfspraulif my frustration levels goes too high, you will hear about it here :-)21:30
wolfspraulxburst-tools just took me 15 months, and it is _almost_ in unstable now21:30
wolfspraulwpwrak: on the 2.6.36 release day, we should have made a nice press release with quote from CEO of Ingenic etc.21:33
wolfspraulso many good opportunities, so much work21:34
wolfspraulI am pretty sure if all I described recently about press releases would have been done, it could have been picked up at quite a few places21:34
wolfspraulmaybe more as 'new Chinese CPU supported' or so21:34
wolfspraul'mobile Chinese CPU finding its way into Linux'21:35
wolfspraulha, to go with the nobel peace prize - 'freedom fighters free mobile Chinese CPU'21:35
wpwrakdon't piss off the rulers of your country ;-)21:36
wolfspraulanyway now it's too late, i think 2.6.36 will be released very soon, and I cannot get enough energy into this right now21:36
wpwrakyeah, don't know how well "refreshing" old news works21:36
wolfspraulnot at all21:37
wolfspraulfacts don't cease to be facts, but news ceases to be news21:37
wpwraktoo bad then. nanowar would have been a candidate21:37
wolfspraulthat's almost like Newton's law in media. don't argue with it.21:37
wpwrakyup. nothing's as old as yesterday's news :)21:37
wolfspraulbtw, journalist friend of mine (non-tech) sent me an email, just ranting a bit about his job21:38
wpwrakhehe :)21:39
wolfspraulhe went to Cuba, for journalistic reasons. In such cases they try to have multiple assignments, some confirmed, some 'maybe' (depending on quality of report), etc.21:39
wolfspraulso he finally got his 10 A4 pages article accepted by a large German newspaper as a 'culture report', and got: 400 EUR!21:39
wolfspraul40 EUR / page21:39
wpwrakcuba is a nice place to be sent to for work21:40
wolfspraulsure you can look at it that way, and it won't be 100% 24/7 work, but still21:40
wolfspraul40 EUR / page for a high quality journalistic report21:40
wolfspraulit's really becoming pocket change21:40
wolfspraulyou can actually only do it on the side of a family vacation soon21:40
wolfspraulrecoup some icecream money21:40
wolfspraulthat's for a large newspaper21:41
wpwrakif you can still ffort a family or vacations, let alone both, they're paying too much :)21:41
wolfsprauland for 40 EUR / page, it needs to be interesting, have good pictures, etc.21:41
wpwrakafford even21:41
wpwrakinteresting how the press became so penniless21:42
wpwrakor stingy21:42
wolfspraulwell I think newspapers are going down fast, and the situation at blogs is even worse21:44
wolfspraulinteresting changes ahead21:44
wolfspraultechcrunch sold itself to AOL because, as the founder of techcrunch said, it was just unsustainable21:45
wolfspraulhe was tired of this, tired of that. just unsustainable.21:46
wolfspraulnow within AOL hen can focus on the journalistic side again (he hopes). :-)21:46
wolfspraulI have my doubts about that, I think the site will be flooded with junk advertisement, but OK, I can understand the tiredness.21:46
wpwrakhmm, don't even know techcrunch22:05
wpwrakgroup-editng blog. wow. tres avantgarde22:06
xiangfuwolfspraul: I think I finish the manual page of fped. and I have uploaded to : http://mentors.debian.net/debian/pool/main/f/fped/22:20
wolfspraulhe, cool!22:20
wolfspraulxiangfu: should I email Jan to ask for help?22:21
wolfspraul"Needs a sponsor: YES"22:21
xiangfuwolfspraul: yes.22:24
kristianpaulThe arbitrary-precision counter counts clock cycles of a frequency22:36
kristianpaulsource that is assumed to be free from drift22:36
kristianpaulwpwrak: so you take the risk and guess that...22:36
wpwrakkristianpaul: well, relatively free from drift. ultimately, ntp is synced to nuclear clocks. but i don't know what they are synced to.22:40
wpwrakkristianpaul: ah, they sync to TAI: http://en.wikipedia.org/wiki/International_Atomic_Time23:18
--- Sat Oct 9 201000:00

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