#qi-hardware IRC log for Thursday, 2010-09-30

SiENcEhey bartbes, textmode told me that you have worked on nLove. I would like to have and test this on dingoo. is it possible?07:48
wolfspraulwpwrak: I'm starting to work on the automated schhist stuff on the Qi projects server08:14
wolfspraulfirst I built debian packages for Kicad plus your patches, http://projects.qi-hardware.com/p/fped/downloads/08:15
wolfspraulnot the best location but the fped Debian package will also show up there so I just put it there for now08:15
wolfspraulthen I installed KiCad+your patches, as well as fped, on our buildhost machine, and made the entire /home/git/repositories available to the buildhost as read-only (via nfs)08:16
wolfspraulnow I want to try to run the schhist scripts on the buildhost08:16
wolfspraulthere is a schhist/Makefile but the paths seem quite hardcoded08:16
wolfspraulI cannot find much documentation either, so I will just start and try to run it see what happens :-)08:17
wolfspraulif anybody is curious the documentation on building KiCad+fped is here http://en.qi-hardware.com/wiki/Server_setup#KiCad08:19
wolfspraulI will also keep updating this once we simplify the process, for example with a full Debian package for fped08:19
wolfspraulTextmode: KiCad is the tool we use for the electrical design (schematics), and layout08:20
wolfspraulbasically there are two serious free tools around, KiCad and geda08:20
wolfspraulwe chose KiCad08:20
wolfspraulTextmode: what I am talking about now is a visualization of schematics changes, directly out of the revision control system08:22
wolfspraulyou can see a demo here: http://downloads.qi-hardware.com/people/werner/schhist/xue/08:22
wolfspraulI want to automate this so that all KiCad projects on the Qi projects server automatically have it08:23
wolfspraulpretty cool stuff I think :-) at least very important for the goals of our copyleft hardware project08:23
wolfspraulthe visualization creates opportunities to join the development or review process08:24
wpwrakwolfspraul: sounds good so far :) most of the hardcoded stuff is meta-data. links to the project page and such.08:56
wpwrakwolfspraul: there are a few tricky/irregular bits, though: projects with more than one kicad project inside, the use of the -f option, and the project name08:57
wpwrakwolfspraul: you need -f if the project doesn't assign useful subsheet names. right now, only SIE CEIMTUN has this problem.08:58
wpwrakwolfspraul: one could of course consider this a bug in the project :)08:58
wpwrakwhat -f does is that it switches to using the file names instead of the subsheet names08:58
wolfspraulwpwrak: yes I see it already09:03
wolfspraulyou try to give it a positive spin though09:03
tuxbrainwolfgang: apart of empower the visibility/participation on existing project I bet including this will atract other project to join qi-hardware :)09:03
wolfsprauleven though almost every project has its own options, at least they are not called -wpan-cntr etc :-)09:03
wolfspraulexporting the git repos via nfs was utter nonsense of course09:04
tuxbrainonly a feeling, but I see some bright eyes when I show the demo on CEDI09:04
wolfspraulwpwrak: do the scripts use git commands to go through the revisions?09:04
wpwrakwolfspraul: (nonsense) why ? you need them, too09:05
wolfspraulthe repos?09:05
wpwrakwolfspraul: yes, it checks out the old revisions. but it does this in a temp directory09:05
wolfspraulyes but I exported the actual git repository folder, which I think is not a good idea09:05
wolfspraulinstead, I just need to do a git clone with the git:// read-only urls09:06
wpwrakwolfspraul: it uses the current repos as a starting point09:06
wpwrakwolfspraul: (repo/clone) is there a difference ? :)09:06
wolfspraulsince the nfs export is read-only, you cannot run git commands in there09:06
wolfspraulyes sure, because you wouldn't want to run the scripts on an nfs export of the actual server-side repository09:07
wolfspraulthat's a bit hard-core09:07
wolfspraulmy bad09:07
wpwrakwolfspraul: hmm, read-only should be okay. not that i've tried it, but the things that run there shouldn't need to write09:07
wolfspraulbut they need to checkout somewhere otherwise how can you run scripts on the files09:07
wpwrakwolfspraul: yes, they get checked out in repo-root/_something directories09:08
wpwrakwolfspraul: or, rather, repo-parent09:08
wolfspraulsure but it's a clone09:08
wpwrakwolfspraul: e.g., if you have ben-wpan in /foo/bar/ben-wpan, then the temp directories are in /foo/bar/_whatever09:08
wolfspraulare the scripts safe to be run from another folder?09:09
wolfsprauldid you ever test it? I see lots of paths in them, so I tend to stay with the 'default'09:09
wpwrakwolfspraul: the temp directories are called _gitsch2ps and _schhist2web09:09
wpwrakwolfspraul: hmm, which paths do you mean ?09:10
wpwrakwolfspraul: the scripts shouldn't care about the directory they run from, as long as they can figure out where the rest of the scripts lives09:10
wpwrakwolfspraul: what you have to tell them is where the checked-out tree is (with .git repo) and where in that tree the kicad project lives09:11
wolfspraulI'll work my way through, no worries09:11
wpwrakwolfspraul: and they need the name of the output directory. that can be a relative path.09:12
wpwrakof course, as always, you may find bugs :)09:12
wpwrakthe huge number of things in schhist/Makefile is mainly because i didn't make a project-name -> environment mapping. things like SCHHIST_HOME_URL and SCHHIST_COMMIT_TEMPLATE are just a function of the project name09:13
wolfspraulsure sure09:14
wolfspraullet me work a bit09:14
wpwraksomething that's highly project-specific is SCHHIST_ORDER. that's the "human-friendly" order of the sheets. not really sure how to automate this.09:15
wolfspraulhow efficient is the script, can I run it every 5 minutes in a cron job going over all KiCad projects?09:15
wpwrak(SCHHIST_ORDER) the same information can exist at other places (other scripts), but there's not really a standard representation for it09:15
wolfspraulor I have some kind of flag that is set whenever a commit is made09:16
wolfspraulbut if the scripts are efficient they will just look at a timestamp and then do nothing anyway09:16
wolfspraulnot sure09:16
wpwrakhmm, 5 minutes may be too often. maybe submit an at job 5 min after the last run terminated09:16
wpwrakif there's a big update, the scripts can run for a long while09:16
wolfspraulsure but that's no problem09:16
wolfspraullet's say there are no changes at all09:17
wpwraklemme see how long they take if there's nothing to do ...09:17
wolfspraulhow efficient is the script in 'doing nothing'?09:17
wolfspraulyes, thanks09:17
wolfspraulif it's not very efficient, I will have a little commit flag09:17
wpwrakit takes a few minutes because the pngs and the index.html are always regenerated. only the pnm and pdfs are cached.09:18
wolfspraulthe thing is I want to run the scripts immediately after a commit, to give people some satisfaction in looking at what they just committed visually09:18
wpwrakyes, i understand09:18
wolfspraulfew minutes, wow09:18
wolfspraulso I cannot run this over all KiCad projects09:18
wolfspraulno problem I'll have a little flag then09:18
wpwrakyou're running right on the machine that will host the output files, right ? i.e., no rsync needed afterwards ?09:20
wolfspraulno it's the buildhost09:21
wolfspraulI think it fits better there, from the load characteristics09:21
wolfspraulI'll figure out how to get the files to the Apache server, there are many options09:22
wolfspraulrsync, nfs, redirects, etc.09:22
wpwrakokay. so, testing the timing with rsync then. let's see how it goes ...09:22
wolfspraulI'm not there yet, step by step09:22
wolfspraulI should probably clone kicad-libs too, right?09:22
wolfspraulis schhist able to deal with kicad-libs, also in revisioning?09:23
wolfspraulprobably not yet :-)09:23
wpwrak6 minutes 11 seconds, with git-pull at the beginning and rsync at the end09:24
wpwrak(that's for all the projects i'm watching, ben-wpan/(atusb, atusd, cntr), xue, sie-cemtun, sie-v209:25
wpwrak(kicad-libs) no, i don't track revisions there. i just use the current state09:26
wpwrakglobal revisions would be interesting :)09:26
wolfspraulso I need to have kicad-libs checked out?09:27
wpwrakyes. and if they use anything else, that too. ideally, just check out all the projects.09:28
wpwrakthings that span projects are a bit of a weakness in the way git is normally used. e.g., in openmoko's svn, we just had one tree for everything. svn lets you check out subtrees and such, so things far away don't get that much in the way.09:30
wpwrakwith git, this is harder. not that the svn approach would be without its own issues. e.g., limiting things to directories doesn't work perfectly.09:31
SiENcE@ Textmode, bartbes does not respond. can i find infos for nLove somewhere else?09:43
TextmodeSiENcE: bartbes *is* nlove atm.09:44
Textmodebut yes, I think he's asleep.09:44
mthmore likely afk but not asleep09:48
wolfspraulwpwrak: hmm09:55
wolfspraulmy folder setup is a bit different, I run into a seg fault09:56
wolfspraulI have a home dir /home/schhist09:56
wolfspraulthe scripts are in /home/schhist/eda-tools/schhist09:56
wolfspraulthe git repos are in /home/schhist/git09:56
wolfspraulthe output dir is /home/schhist/output09:56
wolfspraulI run the scripts from /home/schhist09:57
wolfspraulthe segfault comes from gitsch2ps, which has a parameter /home/schhist/_schhist2web09:57
wolfspraulthis directory does not exist at the time gitsch2ps is called. when I create it before running schhist2web, it is deleted before gitsch2ps is called09:58
wolfspraulstill investigating09:58
wolfspraulmaybe I should run everything in exactly the same folder structure you have09:58
B_LizzardHi, I'm looking to build a battery indicator type thing10:08
B_LizzardI see there exists a file called /sys/class/power_supply/battery/capacity but I was wondering what are the values it emits10:09
B_LizzardIt never reaches 100%10:09
B_Lizzardkernel bug?10:09
larscna, i would say thats a natural thing for batteries10:10
larscescpecialy if such simple methods of measuring it are used10:11
B_LizzardAh, OK10:11
B_LizzardI suspect things will be better in the future10:11
larscthe driver is pretty dump. it meassures the current battery voltage and compares it to a reference voltage which is supposed to be 100%10:12
wolfspraulwpwrak: the segfault is from eeschema, I will rebuild kicad on the buildhost10:14
B_LizzardI see.10:20
wpwrakwolfspraul: how do you invoke schhist ?10:20
B_LizzardI suppose things should be even harder if the system is to handle nokia batteries or worse, knockoff nokia batteries10:21
larscsince it probably has a different reference voltage10:21
wolfspraulwpwrak: from a Makefile, as user schhist10:22
B_LizzardI think Nokia batteries have internal circuits which give a better estimate10:22
B_LizzardAre those supported?10:22
B_LizzardOr is support planned?10:23
wpwrakwolfspraul: i mean, with which arguments ?10:23
larscB_Lizzard: nope. there is no method to access those10:23
B_LizzardOK, thanks a lot, larsc.10:23
larscB_Lizzard: but you could probably build a hw hack to do so10:23
wolfspraulyou mean schhist2web? exactly like you do in Makefile10:24
wolfspraullet me rebuild kicad first, maybe that solves the problem already10:24
wolfspraulthe scripts are a bit hard to follow/debug for a newbie, but I'll find my way through10:25
wolfspraulI just wanted to keep you posted10:25
wpwrakwolfspraul: hmm, ben-wpan would be in /home/schhist/git/ben-wpan/ or /home/schhist/ben-wpan/ ?10:26
wpwrakwolfspraul: (hard to debug) yeah, i know :)10:26
wolfspraulI can change any of this to make it work, and I will :-)10:27
wpwrakwolfspraul: and you're running schhist2web from /home/schhist/eda-tools/schhist/ ? with, say, ./schhist2web ../../ben-wpan atrf/wpan-atrf.sch _atusb   ?10:27
wolfspraulfirst I ran from /home/schhist10:28
wpwrakwolfspraul: okay. so what does the line look like ?10:28
wolfspraulas eda-tools/schhist/schhist2web git/ben-wpan atrf/wpan-atrf.sch output/atusb10:28
wpwrakokay, perfect10:28
wolfspraulhmm, there is a _gitsch2ps folder in /home/schhist/git10:36
wolfspraulI'm a bit worried if my folder structure is different from yours the various scripts will not hand over stuff correctly, look in the wrong place10:37
bartbesSiENcE: of course10:37
bartbeswhat do you want?10:37
wpwrak_gitsch2ps is where it should be10:38
wolfspraulI will continue tomorrow, late here...10:38
wpwrakit needs to be there so that eeschema will find things in other projects that are accessed via relative paths10:38
wpwrake.g., ../kicad-libs/etc10:39
SiENcEhey bartbes,10:40
SiENcEwhat is currently done and where is the repository?10:40
SiENcEi wanna build it for dingoo10:40
bartbesSiENcE: well, there kind of is no repo atm10:51
bartbesthere is a downloadable source though10:51
bartbeshere: http://dl.dropbox.com/u/440010/nlove/nlove-0.0.2.tar.gz10:51
SiENcEwell...i can add it to my svn (its open)10:54
SiENcEor you create a github for example10:54
SiENcEwhat is on todo?10:55
bartbesfix audio10:57
bartbesthere is no need for you to host the code though..10:57
bartbeswell, more like, I need to port it11:00
bartbesfrom OpenAL11:00
bartbesand I miserably failed11:00
bartbes(so far(11:00
SiENcEmh maybe i can help you. when i ported GemRB for Dingux ....we also ported thier OpenAL Plugin to SDL_Mixer11:06
SiENcEdoes nLove has any dependencies?11:07
SiENcEbartbes, how do i build this using cmake?11:16
bartbescmake .11:17
bartbesthen you get a Makefile11:17
SiENcEmh i know but11:17
SiENcEis lua included?11:17
SiENcEwhat else is needed?11:17
bartbesyou might want to take a look at the owrt makefile11:18
bartbesdo you need me to link it?11:18
SiENcEyes please11:18
SiENcEwhat lua version should i take?11:22
bartbesSiENcE: so, any progress?11:54
SiENcEmh no. i have problems translation lua11:56
SiENcEwell ...any problem is also a chance ;) ...so no problems here11:59
SiENcEbartbes, what physfs is required?12:11
SiENcEhow do i a crosscompile12:11
SiENcEthe cmake doesn't recognice my mipsel compiler12:12
SiENcEbut only my nativ one12:12
SiENcEnot mipsel-linux-uclibc-gcc12:12
SiENcEi can set SET(CMAKE_C_COMPILER mipsel-linux-uclibc-gcc)12:12
SiENcESET(CMAKE_CXX_COMPILER mipsel-linux-uclibc-g++)12:12
SiENcEbut does not work12:12
TextmodephysFS is a virtual filesystem, its also the heart of löve's achive (and thus, lovechive) support.12:55
kyakrecently, i found a great use for Ben as a movie player in my car12:58
kyaki connect it to audio input and enjoy tvshows while wasting my time in traffic jams12:59
kyakand traffic jams has become horrible now that summer is finished12:59
bartbesTextmode: when he returns, please do tell him that the Makefile contains everything needed to build13:27
bartbes(I will be leaving again soon)13:27
bartbesso yeah, if he just follows the makefile..13:27
TextmodeI'll try13:29
rafakyak: can you play ogv?14:32
--- Fri Oct 1 201000:00

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