--- Day changed Sat Jun 09 2007 00:47 -!- Igor2 [~igor2@catv-5062cea3.catv.broadband.hu] has joined #geda 01:26 -!- bert [~bert@a82-93-149-85.adsl.xs4all.nl] has joined #geda 01:27 < Igor2> hi bert :) 01:30 < Dan> it must be morning for you guys 01:30 < Dan> time for me to pass out from lack of sleep.... 01:30 -!- Dan [~mcmahill@c-67-191-147-129.hsd1.ga.comcast.net] has quit [Quit: ircII+tkirc2] 01:49 < bert> Hi Igor :) 03:12 -!- werner2101 [~Werner@p57B30689.dip0.t-ipconnect.de] has joined #geda 03:20 -!- ivan [~ivan@sdgate2.cvjetno.sczg.hr] has joined #geda 03:20 < ivan> hi all 03:22 < Igor2> hi 03:27 < werner2101> Good Morning everybody 03:34 -!- peterbrett [~user@ptbb2b.girton.cam.ac.uk] has joined #geda 03:39 -!- peterbrett [~user@ptbb2b.girton.cam.ac.uk] has quit [Remote host closed the connection] 03:39 -!- peterbrett [~user@ptbb2b.girton.cam.ac.uk] has joined #geda 03:51 < bert> Good morning everyone :) 03:52 < ivan> hi bert 03:52 < bert> Hi Ivan, how are you ? 03:53 < ivan> fine, I'm really excited about this code sprint 03:53 < werner2101> hi all, what are your plans for today 03:54 < bert> Me too, I hardly can't wait, so I started doing some coding on my projects 03:55 < ivan> I plan to get my gschem_atexit patch merged and also fix recent files code and dialog pos. code to use that 03:55 < bert> I want to add some doxygen comments to pcb, just like Stuart (I think it was) announced the last code sprint :) 03:56 < ivan> I would also like if people tested the new visual feedback for keyboard commands feature 03:57 < werner2101> I'm going to write an import filter for ngspice, written in python, using numpy and maybe pytables/hdf5 04:01 < werner2101> This maybe a small step towards Draft4 in http://geda.seul.org/wiki/geda:data_plotting_improvements 04:04 < bert> Due to family activities I will join in the afternoon/evening approx. 18.00 hours GMT ... ~ 12.00 hourst ET ?? 04:04 < Igor2> >peterbrett> tried to reproduce the gschem compilation problem but now as i didn't use the central configure/makefile but ran make distclean, ./configure, make, make install in libgeda and gschem, it worked 04:04 < Igor2> i guess it's a typical heisenbug :) 04:04 < peterbrett> Hi 04:05 < Igor2> hi :) 04:05 < bert> Hi PeterB :) 04:05 < ivan> hi Peter 04:05 < peterbrett> I'm going to have breakfast, then fix my website, then move Scheme files around for a while until Ales/PeterC arrive, then work on the git repo 04:05 < peterbrett> bbiab 04:16 < Igor2> ohh,and i forgot to set prefix to /usr\ 04:16 < Igor2> s/\\// 04:17 * bert is away: I'm busy doing something 04:46 < Igor2> finally, gschem works :) 04:54 < bert> Hi Igor, what is a heisenbug ? 04:55 < Igor2> !jargon heisenbug 04:55 < jmp> >Igor2> heisenbug /hi:'zen-buhg/ n. [from Heisenberg's Uncertainty Principle in quantum physics] A bug that disappears or alters its behavior when one attempts to probe or isolate it. (This usage is not even particularly fanciful; the use of a debugger sometimes alters a program's operating environment significantly enough that buggy code, such as that which relies on the values of uninitialized memory, behaves quite differently.) Antonym of {Bohr b 04:55 < Igor2> this one :) 04:55 < peterbrett> heh 04:55 < peterbrett> good definition 04:55 < Igor2> jargon file rulez :) 04:57 < bert> Ahh, I see 04:59 < bert> Ok, I just put together a doxygen config file for pcb, and it seems to work properly 05:00 < bert> It generates docs for all funcs, vars and structs, that is a list without any comments since these are not yet in doxygen format 05:01 < peterbrett> Igor2: Did you manage to find a version of git which supports git-clone --depth? 05:01 < bert> It even generates a todo list and an alphabetical index list 05:01 < peterbrett> And a bug list, if you use \bug commands 05:02 < Igor2> >peterbrett> not yet :) 05:02 < peterbrett> Hmm. I recommend removing the packaged version and installing a release from source. 05:02 < Igor2> >peterbrett> i am not sure yet how i would solve it, since i don't want to upgrade my debian from stable to unstable and while i don't know enough git compiling it from source doesn't seem to be a good idea either :) 05:03 < peterbrett> Um, it didn't even have a configure script last time I checked. 05:03 < peterbrett> :) :P 05:04 < Igor2> :) 05:10 < bert> PeterB: where can I send a doxygen config file to contribute to pcb ? SF tracker ? 05:11 < peterbrett> bert: Yep 05:12 < peterbrett> bert: and/or e-mail to list with [PATCH] in subject, but I think tracker is preferred. 05:12 < Igor2> hmm,nice traffic on the mailing list about git and trac :) 05:12 < bert> Patch of Feature request ? 05:12 * peterbrett not a huge trac fan 05:12 < peterbrett> bert: Patch 05:12 < Igor2> neither am i 05:12 < Igor2> i like the basic idea 05:12 < Igor2> but i dislike the following things: 05:12 < peterbrett> it's a really good idea 05:13 < peterbrett> brb 05:13 < Igor2> - i don't believe everythink should be on web primary,so as long as web is a frontend to a sane backend, it's ok, but when there's only web frontend to a service, i tend to dislike it 05:13 -!- peterbrett [~user@ptbb2b.girton.cam.ac.uk] has quit [Remote host closed the connection] 05:13 < Igor2> - i never liked wiki for many reasons and trac seems to be linked to wiki on several levels 05:13 < Igor2> - i find the tracker comment-list uncomfortable to read :) 05:14 < ivan> i find sf trackers hard to use and very inconvenient 05:14 < ivan> the UI is a disaster 05:15 < ivan> not to mention that it's slow 05:15 < Igor2> yup 05:15 < Igor2> and i think this is true for the whole sf, not just for trackers 05:15 < Igor2> but this why i dislike something that has web-itnerface-only :) 05:15 < ivan> me too 05:16 < Igor2> i think it's reasonable to provide some computer-readable interface and web interface so windows users could use their ie and everyone else can develope/download/use sane clients ;) 05:16 < ivan> lol 05:21 < bert> PeterB: doxygen config file is ti SF tracker 05:22 < bert> s/ti/in the/ 05:22 < bert> I will be away for a couple of hours, bye 05:22 * bert is away: I'm busy doing something 05:23 -!- peterbrett [~user@ptbb2b.girton.cam.ac.uk] has joined #geda 05:23 < peterbrett> hi again 05:23 < Igor2> wb :) 05:23 < peterbrett> no sign of pcjc2, I suppose? 05:23 < Igor2> haven't seen him today 05:24 < Igor2> you are never online in the same time,hehe :) 05:24 < peterbrett> lol 05:24 < peterbrett> we are sometimes :) 05:24 < Igor2> i know, i know! split personality! :) 06:08 < ivan> regarding doxygen comments... 06:09 < ivan> it doesn't look nice when i repeat the same description under the \brief and \par sections 06:09 < peterbrett> Ah 06:09 < peterbrett> Okay 06:09 < peterbrett> Maybe I was wrong :-/ 06:10 < ivan> is there a way to tell doxygen to include \brief section automatically? 06:10 < peterbrett> Not sure 06:10 < peterbrett> I'm not that familiar with doxygen 06:11 < ivan> hm... I'll take a look when I finish the patches ... 06:56 < peterbrett> Gosh, it's really quiet around here today 06:56 < peterbrett> And still no sign of PeterC 06:56 * peterbrett wonders what's happened to him 07:08 < Ales> morning folks 07:08 < Igor2> hi ales :) 07:08 < Ales> I just got up, so I am not in the right physical location. :) 07:08 < Ales> hi Igor 07:09 < Igor2> code sprint today? 07:09 < Ales> yup 07:11 < peterbrett> Hey Ales 07:11 < peterbrett> I've been working on reshuffling Scheme files between symbols & libgeda 07:12 < peterbrett> Also trying to get rid of configure substitutions from as many as possible 07:12 < Ales> oh yeah the config.scm file 07:12 < Ales> that'll be nice 07:12 < peterbrett> pcjc2 is just getting out of bed ;) 07:12 < Ales> ah 07:12 < Ales> I have to take my gf to the airport and then I'll head up to DJs 07:14 -!- pcjc2 [~pcjc2@AC9FC5A4.ipt.aol.com] has joined #geda 07:14 < pcjc2> morning everyone! 07:14 < Igor2> hi 07:17 < Ales> hi Peter3 07:17 < Ales> oops... Peter2 07:19 < peterbrett> lol 07:19 < Ales> I don't know if I could handle another Peter. :) 07:19 < peterbrett> Ales: Actually I'm Peter2, he's Peter1 07:19 < Ales> whatever :) 07:19 < peterbrett> I got a puncture... hopefully it's fixed now :/ 07:21 < Ales> in a tire I hope 07:21 < peterbrett> Ales: Should Scheme libraries go in $GEDADATADIR/scheme, or $GEDARCDIR/scheme, or both? 07:21 < peterbrett> Ales: In a tyre, yes :) 07:22 < Ales> in whatever path gschem.scm or gnetlist.scm get installed 07:22 < peterbrett> jawohl 07:22 < Ales> I think it's the DATADIR 07:22 < Ales> cause GEDARCDIR gets redirected to /etc for debian 07:23 < peterbrett> okay 07:23 -!- cnieves [~cnieves@89.pool85-57-71.dynamic.orange.es] has joined #geda 07:23 < cnieves> hi all 07:23 < Igor2> hi :) 07:23 < peterbrett> Hmm 07:23 < peterbrett> Where should config.scm go ? 07:25 < peterbrett> I'm leaning towards GEDADATADIR 07:25 < peterbrett> the variables it defines can always be overridden in system-gafrc 07:26 < Ales> yeah, DATADIR sounds right 07:26 < Ales> okay, I'm going to eat breakfast and I'll be back around 10am EST 07:26 < Ales> later 07:27 < pcjc2> what time is it now EST? 07:27 < Igor2> *** CTCP TIME reply from Ales: Sat Jun 9 07:27:19 2007 07:27 < Igor2> in case EST is local time for him... :) 07:27 < pcjc2> ok, thanks 07:33 < peterbrett> Failure to fix puncture... second attempt in progress :( 07:36 -!- Werner [~Werner@p57B32CE1.dip0.t-ipconnect.de] has joined #geda 07:45 -!- pcjc3 [~pcjc2@AC9FC5A4.ipt.aol.com] has joined #geda 07:45 < pcjc3> now I'm pcjc3 - changed my net connection! 07:46 -!- werner2101 [~Werner@p57B30689.dip0.t-ipconnect.de] has quit [Ping timeout: 620 seconds] 07:47 < peterbrett> pcjc3: Bike fixed, on my way 07:47 < peterbrett> Speak later, all 07:47 < pcjc3> cool, see you in a bit 07:47 < cnieves> ok, see you! 07:52 -!- pcjc2 [~pcjc2@AC9FC5A4.ipt.aol.com] has quit [Ping timeout: 608 seconds] 07:53 < ivan> Ales, what do you intend to do with the gschem-atexit patch? 07:53 < pcjc3> Ales went for breakfast 07:53 < pcjc3> \nick pcjc2 07:53 -!- pcjc3 is now known as pcjc2 07:54 < ivan> ah... 07:54 < ivan> ok, I'll wait until he gets back 07:54 < pcjc2> Ivan: good to see you on IRC 07:55 < ivan> pcjc2: well it's good to see people working together... 07:55 < ivan> pcjc2: this one seems to be a bit quiet, though 07:56 < pcjc2> it hasn't started yet 07:56 < pcjc2> its about 8am in EST 07:56 < pcjc2> Mostly us Europeans about 07:56 < ivan> right, here is 2pm :) 07:56 < pcjc2> 1pm here 07:57 < ivan> pcjc2: what are your plans for today? 07:57 < pcjc2> non descript 07:57 < pcjc2> Ales may want help playing with the new git stuff 07:58 -!- peterbrett [~user@ptbb2b.girton.cam.ac.uk] has quit [Ping timeout: 608 seconds] 07:58 < pcjc2> I'm just currently tinkering with a fix for initialising gthreads 07:59 < ivan> what about your footprint selector and other goodies you mentioned on the dev list? I'm very interested in those things 07:59 < pcjc2> well - I ran into a bit of a design block on those 07:59 < pcjc2> but I could continue coding the GUI elements I guess 08:00 < pcjc2> I'm a bit stuck what the format of the configuration will be - hence how to populate all the auto-complete drop downs and validators etc.. 08:00 < ivan> let me know if I can be of any help 08:01 < pcjc2> The idea is to have it customisable, so various procedures / data can be used to define different "required" or "suggested" attributes etc... for different components / footprints 08:02 < ivan> sounds cool 08:02 < pcjc2> the practicalities are difficult 08:02 < pcjc2> I thought it might fix things to refer to scheme code from the schematic 08:02 < pcjc2> sorry - from the symbols 08:02 < ivan> yes, I saw the post on the list 08:02 < pcjc2> so, for example - a resistor symbol might refer to "resistor.scm" which has lots of resistor specific stuff in it 08:03 < ivan> the idea is intriguing 08:03 < pcjc2> but another way would be to embed data, to avoid it being too programmable 08:03 < ivan> you could use it for other things as well 08:03 < ivan> well, better to be careful wrt security 08:04 < pcjc2> another option would be to "sign" the symbols some how, so it only executes code on ones we ship - without user approval 08:04 < pcjc2> seems a bit heavyweight, but could be done 08:04 < ivan> while reading the thread, I got an idea: how about using cryptographic signing of all symbols shipped with geda 08:04 < pcjc2> snap! 08:04 < ivan> hehe :) 08:05 < ivan> this should solve the security problem 08:06 < pcjc2> I'm not sure if there are still many / any export restrictions on Crypto from the US - we wouldn't want to introduce anything which many users can't get hold of 08:06 < pcjc2> We could just restrict the search paths 08:06 < pcjc2> forbid execution of any referenced .scm code which doesn't fall into our pre-defined paths 08:06 < ivan> possible, but not very flexible 08:07 < pcjc2> then if the user gets a symbol off someone, they explicitly have to copy the .scm file into the correct path for it to execute the code 08:08 < ivan> yes, but that would require root in most cases which is not an option in many places 08:08 < Igor2> or 08:08 < Igor2> you would use a list of approved paths 08:08 < Igor2> and the user could add new paths in the list 08:09 < Igor2> anyway with paths, you need to take care about a few things 08:09 < Igor2> like ..s in the path 08:09 < Igor2> or symlinks 08:09 < Igor2> maybe even the file owner 08:11 < pcjc2> yep - I'm not a fan of paths, and don't know enough to beat the determined cracker 08:12 < Igor2> signitures are not 100% safe either :) 08:14 < Igor2> btw, why do we need to run code for symbols? 08:18 < ivan> as pcjc2 said it: The idea is to have it customisable, so various procedures / data can be used to define different "required" or "suggested" attributes etc... for different components / footprints 08:18 < Igor2> well 08:18 < Igor2> if you want it really safe 08:19 < Igor2> maybe you shouldn't use scheme 08:19 < Igor2> maybe you should use an interpreter written for this task which knows only a very limited set of operations that can not be abused to crack the host comp 08:20 < ivan> this would be an overkill, imho 08:20 < Igor2> if it is really only about required and suggested languages 08:20 -!- peterbrett [~user@AC9FC5A4.ipt.aol.com] has joined #geda 08:20 < Igor2> s/languages/attributes/ 08:20 < peterbrett> hello again 08:20 < ivan> hello 08:20 < Igor2> then the language would have only 4 or 5 commands... :) 08:20 < pcjc2> The slight problem, is the flexibility is good 08:20 < Igor2> (or i miss something important:) 08:21 < pcjc2> what would be nicest is a sandbox 08:21 < ivan> yes 08:21 < Igor2> some language supports that, but... 08:21 < Igor2> it reminds me on js 08:21 < pcjc2> I've no idea if guile does 08:21 < Igor2> for flexible client side whatevers for homepages 08:21 < pcjc2> so we get back to being able to trust the code we execute 08:21 < Igor2> people believed js was so safe that it couldn't be used to abuse the host computer trough the browser 08:22 < Igor2> the first script i've seen that downloaded an exe and executed it without asking the user was less than 100 lines long ;) 08:22 < pcjc2> hehe 08:22 < pcjc2> so not properly sandboxed 08:22 < Igor2> yup 08:23 < Igor2> and here comes complexity 08:23 < Igor2> the more flexibility you want, the more complex your language will grow 08:23 < pcjc2> its difficult to ensure for a flexible language which doesn't support it properly 08:23 < Igor2> and the less probable your sandbox willreally protect you 08:23 < Igor2> and soon it will be as risky as trying to protect the path ;) 08:23 < peterbrett> emacs users: http://www.emacswiki.org/cgi-bin/wiki/Git 08:23 < Igor2> i've written an interpreter 2 summers ago 08:24 < Igor2> and i made a sandbox stuff for it 08:24 < Igor2> i found that the best was to turn off _all_ io operations :> 08:24 < ivan> :) 08:24 < Igor2> (and dynamic loading of new functions from .so files of course:) 08:25 < pcjc2> Off to lunch for a bit - will be about though 08:36 -!- peterbrett [~user@AC9FC5A4.ipt.aol.com] has quit [Ping timeout: 608 seconds] 09:11 -!- peterbrett [~user@AC9FC5A4.ipt.aol.com] has joined #geda 09:11 -!- al [~al@74.94.229.218] has joined #geda 09:11 < peterbrett> hey 09:17 < pcjc2> Hi Carlos: thanks for applying that patch 09:17 < pcjc2> I didn't have time to test it against guile 1.6 09:17 < pcjc2> (the main reason I was holding off) 09:18 < cnieves> hi Peter: thank you for providing it! 09:18 < pcjc2> Do you have guile1.6 to test it against? 09:18 < cnieves> it won't hurt guile 1.6 09:18 < pcjc2> was the bug specific to 1.8? 09:18 < cnieves> I think so 09:19 < cnieves> nil was provided just for compatibility in guile 09:19 < cnieves> it seems they removed nil in 1.8, and that's the reason guile 1.8 doesn't know about it. 09:20 < cnieves> in fact, nil usage is not recommended at all. Use #f or () ,as you said 09:20 < pcjc2> I got that of a mailing list archive somewhere 09:20 < cnieves> peterbrett: I'm looking at patch [ 1733728 ] Fix for lack of permission-related file attributes on MinGW 09:21 < cnieves> peterbrett: I remember one of your mails said you have some notes about compiling under windows. Do you agree with the mentioned patch? 09:22 < pcjc2> It was probably me 09:22 < cnieves> pcjc2: I remember I read it on some webpage some time ago. 09:22 < peterbrett> cnieves: I haven't touched windows in months for anything other than gaming, haven't got a clue sorry :-/ 09:22 < peterbrett> cnieves: I think it was pcjc2 09:23 < cnieves> ok, sorry! 09:24 -!- peterbrett [~user@AC9FC5A4.ipt.aol.com] has quit [Remote host closed the connection] 09:24 -!- peterbre` [~user@AC9FC5A4.ipt.aol.com] has joined #geda 09:24 -!- peterbre` is now known as peterbrett` 09:24 < pcjc2> http://www2.eng.cam.ac.uk/~pcjc2/geda/fix_win32.diff 09:24 < pcjc2> This is a summary of the _hacks_ needed to make windows work 09:25 < pcjc2> Caesar was working up proper patches with feature based 09:25 < pcjc2> #ifdef workarounds 09:28 < cnieves> ok. you commented out everything about file permissions. I think Cesar's patch is better in this case. 09:31 < pcjc2> yep - as I said, it was hacks to see what areas needed work to make it compile and run 09:32 < pcjc2> There are some things which might be useful in my patch 09:33 < pcjc2> in o_picture.c 09:33 < pcjc2> -#include +#include 09:33 < pcjc2> and the bits which avoid "basename" 09:34 < peterbrett`> Argh, we shouldn't be using gh at all! 09:41 -!- cnieves [~cnieves@89.pool85-57-71.dynamic.orange.es] has quit [Ping timeout: 620 seconds] 09:49 < pcjc2> gah - thats the second time I've lost work on configure.ac 09:49 < pcjc2> forgetting its actually configure.ac.in I should edit!! - DOH 09:55 -!- Dan [~mcmahill@c-67-191-147-129.hsd1.ga.comcast.net] has joined #geda 09:56 < Dan> hello 09:56 < pcjc2> hi 09:56 < ivan> hi 09:56 < Igor2> hi 09:59 < ivan> does anyone know why git-send-email shortens Subject: lines? 10:00 < peterbrett`> ivan: Probably to make you write short & succinct log message header lines? 10:00 < peterbrett`> (I don't actually know) 10:02 < ivan> hm... it seems it stops at the first newline 10:02 < pcjc2> yep 10:02 < peterbrett`> ivan: Yep 10:02 < pcjc2> thats a known issue 10:02 < peterbrett`> it's supposed to 10:03 < pcjc2> Makes it really evil for rebasing stuff based on CVS, as it uses the email formatting programs to do it 10:03 < pcjc2> and hence all the commit messages get munged. I've got a hacked version of a few of the git tools which avoid this when rebaseing 10:03 < peterbrett`> All of the git tools -- ALL of them -- assume you write your log messages in the form 10:03 < peterbrett`> pcjc2: Or you could just write your log messages in the right format 10:04 < pcjc2> Thats a bug IMHO, but I guess its defined 10:05 < ivan> but sometimes the log message is exactly a bit longer than the short one-liner and too short for the long description 10:05 < ivan> oh well... 10:05 < peterbrett`> ivan: Just make it shorter then. 10:05 < peterbrett`> "Fix Bug #24834438" is a perfectly valid patch summary, as long as you write a longer log message below 10:08 < Ales> okay, I'm alive 10:09 < ivan> hooray :) 10:09 < Ales> now, I want to commit some stuff first 10:10 < Ales> starting with the visual key thingy from Ivan 10:11 < pcjc2> I've got a patch for the gthread initialisation - I'll pastebin it when its ready 10:11 < Ales> cool, by all means checkin as much as possible 10:11 < pcjc2> Just wanted to run it past people and see its ok 10:12 < ivan> Ales, what about gschem-atexit patch? 10:12 < Ales> I'll do that one next. :) 10:12 < ivan> great 10:12 < Ales> then I'll start working on git 10:13 < peterbrett`> Ales: Using a separate config.scm isn't going to work :( 10:13 < peterbrett`> Ales: the rc files can't find it without knowing its contents already 10:13 < peterbrett`> It's a chicken & egg problem :) 10:13 < Dan> Ales, you up in NH? 10:15 < Dan> anyone know if the last geda snapshot works with guile-1.8.1 or if it is just the configure test which is broken? 10:16 < pcjc2> http://www.pastebin.ca/552875 10:16 < peterbrett`> Dan: Elaborate? 10:16 < ivan> i've just sent 3 patches to the dev list 10:17 < Dan> checking for guile-config... (cached) /usr/pkg/bin/guile-config 10:17 < Dan> checking if scm_eval_x() takes a second module argument... no 10:17 < Dan> checking for scm_make_smob_type in -lguile... no 10:17 < Dan> configure: error: Upgrade Guile, At least version >= 1.3.2 10:17 < Dan> *** Error code 1 10:17 < Dan> I haven't dug further yet 10:18 < Dan> I'm building all packages in NetBSD's pkgsrc which use guile so I can see which ones will have immediate build problems with 1.8.1 before I commit the new guile package 10:18 < Ales> Dan: yes, I'm in NH at DJs place! 10:20 < ivan> Ales, please check the mailing list 10:20 < Ales> Dan: the latest geda snapshot should build/run fine with guile 1.8.x 10:20 < Dan> ok. there may be a bug in my guile then 10:20 < Dan> after this (very very long) build finishes I'll look into what happened 10:20 < Dan> btw, that numbers.test failure has been fixed in guile cvs 10:21 < Dan> and I've gotten some help in debugging the other failures 10:21 < Ales> should I just apply all of them in the order they were sent? 10:21 < ivan> yes 10:21 < Ales> and I take it the middle one superceeds the previous gschem_atexit patch? 10:21 < ivan> no, the middle one is on top of it 10:22 < ivan> the first is the original gschem-atexit patch 10:22 < Ales> okay, so this is the complete set 10:22 < ivan> yes 10:23 -!- dj-mon [~dj@lust.delorie.com] has joined #geda 10:23 < pcjc2> Ales - do those g_therad_init calls look ok? 10:23 -!- dj-mon is now known as dj 10:24 < Ales> is there a patch for them somewhere? 10:24 < pcjc2> http://www.pastebin.ca/552875 10:26 < Dan> hi dj 10:26 < dj> Hi folks! 10:26 < Ales> what version of gtk+ required threading for the file selector? 10:26 < Dan> want me to email you a .pcb file which shows the connection scanning bug? 10:27 < pcjc2> I'll have a dig, but I believe its been like that for some time 10:27 < peterbrett`> In my libgeda config.h: #define GEDARCDIR "none" 10:27 < pcjc2> its just the verbose warning which was added recently 10:27 < peterbrett`> I think this is the Wrong Thing? 10:27 < dj> yes please 10:29 < pcjc2> hmm - might not be GTK which is using threads - I get a feeling it might be GnomeVFS 10:29 < Dan> sent 10:29 < Ales> it's correct cause in g_rc.c it checked and if it isn't none something else happens 10:29 < peterbrett`> Hmm 10:29 -!- cnieves [~cnieves@104.pool85-57-64.dynamic.orange.es] has joined #geda 10:29 < Ales> my only concern with calling g_thread_init on systems where it isn't needed 10:29 < cnieves> hi all (again) 10:30 < peterbrett`> Ales: I want to get libgeda to define the appropriate Guile symbols in g_register 10:30 < peterbrett`> How do you think I should do that? 10:30 < Ales> I haven't ever thought about that. 10:31 < pcjc2> Ales: http://www.nabble.com/bugs-regarding-late-g_thread_init()-calls-t2907988.html 10:32 < peterbrett`> Okay, I'll define the Guile symbol rc-path to be the result of g_rc_parse_path(). 10:32 < Ales> peter: which rc vars do you want to remove from the .scm files? 10:33 < peterbrett`> I want to be able to have data-path, rc-path & path-sep 10:33 < peterbrett`> Defined as Guile symbols 10:34 < peterbrett`> Anyway, I'll bbiab 10:34 < Ales> some of those are set depending on ./configure or environment variables 10:34 < peterbrett`> Yep 10:34 < peterbrett`> Speak later (~20 min, hopefully) 10:35 < peterbrett`> bye 10:41 < Ales> peterC: yeah, go ahead and commit those threading patches 10:41 < Ales> it's probably okay to call g_thread_init 10:43 < pcjc2> ok, will just write up a changelog 10:44 < Ales> Ivan: your key display patch has been applied 10:44 < ivan> thanks 10:44 < pcjc2> just now? 10:45 -!- peterbrett` [~user@AC9FC5A4.ipt.aol.com] has quit [Ping timeout: 620 seconds] 10:46 < Ales> just now. 10:46 < Ales> you may have to update... sorry! 10:46 < pcjc2> done, no problem 10:47 < ivan> I think once the git repo is set up, all this will become *much* easier... 10:48 < cnieves> pcjc2: tell me when you commit. I have one more patch ready. 10:50 < Ales> Ivan: are you sure that your patches include all the necessary code for gschem_atexit? 10:50 < Ales> the three that you just sent me 10:50 < ivan> yes, I checked several times 10:50 < bert> Hi all :) 10:50 < Ales> okay, I'll just build it. :) 10:50 < cnieves> are those the same ones you sent to the mailing list? 10:50 < cnieves> hi Bert 10:51 < bert> Hi Carlos 10:51 < ivan> yes 10:52 < Ales> Ivan: okay I found the gschem_atexit code 10:53 < Ales> rebuilding now 10:54 -!- Cesar [~Cesar@c9342d24.virtua.com.br] has joined #geda 10:54 < pcjc2> Carlos: Committed just now 10:54 < cnieves> ok, updating 10:55 < Cesar> Hi, everybody. 10:55 < Ales> let me know when you are done peter, cause I have Ivan's patches applied 10:55 < pcjc2> its done, thanks 10:55 < Ales> okay rather, Carlos, when you are done 10:56 < Ales> Hi Cesar 10:58 < pcjc2> Hi Cesar - Carlos has been looking at committing some of the Win32 fixups you were working on 10:59 < cnieves> Ales: done 10:59 < cnieves> Cesar: I applied your button order patch. 10:59 < Ales> okay, I'm still building, so if anybody else needs to commit go ahead 11:00 < Cesar> cnieves: Thanks, I will test it right away. 11:00 < cnieves> Cesar: I was looking at the file permissions patch. BTW do you have cvs write access? 11:01 < Cesar> cnieves: No, i do not have cvs write access yet. 11:03 < cnieves> Cesar: I think pcjc2 changed g_mkdir call by mkdir. Does g_mkdir exist in mingw? 11:03 < pcjc2> glib ought to be portable 11:03 < cnieves> I thought so, then why did you change it? 11:03 < Ales> yeah using glib funcs is fine 11:03 < pcjc2> urm - I wasn't aware that I had 11:04 < cnieves> ok 11:04 < ivan> g_mkdir showed up in glib 2.6 11:04 < Cesar> Problem is, g_mkdir has a second permission argument. 11:05 < Cesar> My solution was to define the missing flags as zero. 11:05 < pcjc2> I see what I did - yet 11:05 < pcjc2> I think Cesar's solution is best 11:05 < pcjc2> although it does expose bugginess in GLibs portability 11:06 < pcjc2> "The mode argument is ignored on Windows." it wouldn't have hurt to define the modes as 0 in glib then 11:07 < cnieves> Cesar: I think your patch is ok. I'll go and apply. 11:07 < pcjc2> one second 11:08 < cnieves> Cesar is defining missing permissions as 0, and they are all used with an OR operator when defining file permissions, so I think is ok. 11:08 < Ales> one more build before I commit all of Ivan's patches 11:08 < pcjc2> I just wanted to look where those defines go 11:10 < cnieves> pcjc2: is it ok? 11:10 < pcjc2> Can we bug Ales for an answer on that 11:10 < Ales> what? 11:10 < Ales> :) 11:10 < pcjc2> the conditional over-rides for modes in f_basic.c 11:10 < pcjc2> is that the "right place", or should they go in config.h somehow 11:11 < pcjc2> +#ifndef S_IWGRP 11:11 < pcjc2> +# define S_IWGRP 0 11:11 < pcjc2> +#endif 11:11 < pcjc2> and the like 11:11 * bert is away: I'm busy doing something else 11:11 < pcjc2> also done in o_misc.c 11:11 < Ales> I'm guessing it is probably not the easiest thing to check in configure? 11:11 < Ales> could you put them instead in a define.h file? 11:12 < Ales> if it isn't straighforward to check for in configure, then it's okay to put them in a toplevel header file (if they are used in more than one place) 11:12 < pcjc2> libgeda/include/defines.h ? 11:12 < Ales> that's fine with me 11:12 < pcjc2> used in libgeda and gschem 11:12 < Ales> I'm about to commit Ivan's patches 11:12 < Ales> perfect place then 11:13 < pcjc2> cool 11:13 -!- peterbrett [~user@ptbb2b.girton.cam.ac.uk] has joined #geda 11:13 < peterbrett> hi again 11:13 < cnieves> ok, I'll change it. 11:13 < cnieves> hi PeterB 11:13 < pcjc2> PeterB: Hi - back in the land of the AC-Supply again ;) 11:13 < peterbrett> :P 11:14 < peterbrett> now... where was I? :P 11:14 < pcjc2> PeterB am about to export xgsch2pcb 11:14 < Ales> commiting... committed. 11:14 < Ales> okay, I'm done patching, I need to work on the git thingy 11:15 < Ales> Ivan: please test the latest in CVS. 11:18 < Ales> okay git.gpleda.org is up and running 11:18 < peterbrett> 403 Forbidden - No projects found 11:19 < Ales> there are no probjects 11:19 < peterbrett> :) 11:19 < peterbrett> Why the 403? 11:19 < Ales> it does that without projects 11:19 < peterbrett> Fair enough :) 11:19 < peterbrett> I think pcjc2 is creating a xgsch2pcb repo as we speak 11:19 < pcjc2> git-cvsimport -v -i -k -p -Z,9 -d :pserver:anoncvs@cvs.seul.org:2401/home/cvspsrv/cvsroot -o master geda/xgsch2pcb 11:20 < pcjc2> is the command line I used, no conversion of author userids to proper name and email addresses 11:20 < Ales> so you guys want xgsch2pcb to be its own git? 11:20 < pcjc2> verbose logging etc... 11:20 < pcjc2> not especially 11:20 < pcjc2> but it wants to be in git somehow 11:20 < peterbrett> Ales: I think it would be a good idea for each distinct project to have separate repos 11:20 < peterbrett> gaf is a project, xgsch2pcb is a project... etc. 11:20 < Ales> okay, so far we have gaf, xgsch2pcb, website 11:21 < peterbrett> Probably enough for the time being :) 11:21 < pcjc2> xgsch2pcb could be a friend of geda 11:21 < pcjc2> (gschem) 11:21 * peterbrett is confused by pcjc2 11:21 < Ales> friend? 11:21 < pcjc2> gaf = gschem and friends 11:21 < peterbrett> lol 11:21 < Ales> ah 11:21 < peterbrett> pcjc2: Stabilise it and make it useful first? 11:21 < Ales> :) 11:21 < pcjc2> I lied earlier - it has no branches 11:22 < pcjc2> it has a tag 11:22 < Ales> okay, so what's the command I should run to do a quick dry run create of the gaf git repo? 11:22 < Ales> ooops... maybe I should install git on seul first 11:22 < pcjc2> dry run as in doesn't do anything ? 11:22 < pcjc2> probably best to make a dir and run a command similar to the one I posted 11:22 < Ales> no, dry run as in create git file so that I can put something in place 11:23 < Ales> okay 11:23 < pcjc2> git-cvsimport -i -k -p -Z,9 -d :pserver:anoncvs@cvs.seul.org:2401/home/cvspsrv/cvsroot -o master geda/gaf 11:23 < pcjc2> is what I use 11:23 < peterbrett> Once you've run the git-cvsimport you'll have a directory called .git 11:23 < pcjc2> which keeps all branch names, apart from HEAD (which is special in git), which it renames to "master" 11:23 < Ales> okay 11:23 < peterbrett> Rename it to xgsch2pcb.git, and put it in the directory where gitweb & git-daemon looks for repositories 11:24 < pcjc2> that command doesn't attempt to guess where merges occurred, and unless the commit messages have the right magic keywords in them, it doesn't work reliably 11:24 < Ales> okay 11:24 < pcjc2> I think PeterB and I came to the conclusion that merge tracking wasn't important for our switchover - the commits are all there etc... and there are no active branches anyway 11:25 < Ales> yeah, I'm okay with that 11:27 < dj> installing software packages on windows really makes you appreciate yum 11:28 < peterbrett> does it ever! 11:28 < ivan> btw, does anyone here use archlinux? pacman just rocks :) 11:29 < Ales> which debian package has git-cvsimport? 11:30 < pcjc2> ah - please ensure you use the newest available cvsps 11:30 < pcjc2> just searching the package out 11:30 < Ales> 2.1 good engouh? 11:30 < pcjc2> git-cvs 11:31 < Ales> okay 11:31 < pcjc2> 2.1 but needs patches 11:31 < pcjc2> IIRC 11:31 < Ales> okay, that's a problem then 11:31 < pcjc2> It might do without, but its best to be consistent 11:31 < peterbrett> pcjc2: It'll probably do okay as long as no-one checks in while export in process 11:31 < peterbrett> Every time I've had a problem, it's because CVS isn't really concurrent 11:31 < Ales> I'll take cvs offline when I do things for real 11:32 < pcjc2> improves tag handling I believe 11:32 < Ales> 2.1 is the latest 11:32 < pcjc2> I can give you the URL of the git repo 11:32 < pcjc2> 2.1 is latest release, but the guy who maintains it is lazy 11:32 < Ales> is this really necessary? 11:32 < peterbrett> Okay, I'm going to make libgeda define "geda-rc-path", "geda-data-path" and "path-sep" as Scheme variables. 11:32 < pcjc2> some other guy has collated various bug fixes in a git-repo to go ontop of 2.1 11:33 < peterbrett> In g_register_libgeda_vars 11:33 < peterbrett> s/lazy/busy/ 11:33 < pcjc2> I've had occasional issues with the older cvsps, and less with the patched version - I might be wrong 11:33 < Ales> lemme try it with 2.1 and see what happens... 11:33 < pcjc2> s/lazy/busy/ of course, sorry - I didn't mean the guy any offence 11:34 < pcjc2> I'll make a log with my cvsps, and we can compare notes 11:34 < pcjc2> (You can run cvsps separeately if you get the command line right, and feed that output manually to git-cvsimport 11:35 < ivan> Ales, I just checked the latest cvs -- all the code is there, works fine... thanks! 11:35 < Ales> this is what I got when when I ran git-cvsimport: 11:35 < Ales> AuthReply: I HATE YOU 11:35 < Ales> maybe I should pick the right username 11:35 < Ales> ah better 11:35 -!- peterbrett [~user@ptbb2b.girton.cam.ac.uk] has quit [Remote host closed the connection] 11:35 -!- peterbrett [~user@ptbb2b.girton.cam.ac.uk] has joined #geda 11:36 < Ales> its running now 11:36 < Ales> doing something 11:36 < pcjc2> eating my words... I'm not using patched cvsps 11:37 < pcjc2> wounder if thats why it dropped some stuff recently ;) 11:37 < peterbrett> lol 11:37 < Ales> how long is it supposed to run fo? 11:37 < Ales> r 11:37 < pcjc2> a long time 11:38 < Ales> Committing initial tree a157fd973b4d818573a45961560464a2ef75e9f4 11:38 < Ales> ooh 11:38 < pcjc2> could be an hour or so I guess 11:38 < Ales> an hour? 11:38 < Ales> goodness 11:38 < pcjc2> you ran with -v? 11:38 < Ales> nope 11:38 < Ales> should I have? 11:38 < pcjc2> ok - boring with no output to watch 11:38 < peterbrett> It should be faster if it's not throttled by the network connection 11:38 < Ales> can I just control c it? 11:38 < pcjc2> you have to delete the .git dir and start again for most safety 11:38 < Ales> okay 11:38 < pcjc2> and possible the cvsps cache in ~/.cvsps/..... 11:39 < Ales> done. 11:40 < Ales> hmm.. get this: 11:40 < Ales> connect error: Network is unreachable 11:40 < Ales> and then it continues on 11:40 < peterbrett> yes, I get that too 11:40 < Ales> odd 11:41 < peterbrett> I've got it for months, it's probably a cvsps bug 11:43 < Ales> verbose is well verbose 11:44 < peterbrett> it's "verbose", not "slightly chatty" 11:44 < pcjc2> how fast is it? 11:44 < peterbrett> Unlike with some tools, where you have to add "-v -v -v --debug" to get any useful debugging output :P 11:44 < pcjc2> mine is slow enough to read the individual fetches 11:44 < Ales> it's scrolling faster than I can really read it 11:44 < pcjc2> good to hear ;) 11:46 < cnieves> Patch #1733826: Avoid writting '\r' to files on MinGW. It changes all fopen(file, "w") to fopen(file, "wb"). However this is not documented in the fopen man page installed on my system. What about yours? 11:47 < pcjc2> 0b22874791baef1fb3626e5046725d61 cvsps.output 11:47 < pcjc2> from my raw run of cvsps 11:48 < Ales> run run run 11:48 < peterbrett> cat .git/refs/heads/master 11:48 < Ales> this is a read only operation on the cvs repsoitory I hope? 11:48 < peterbrett> both of you :) 11:48 < peterbrett> Ales: yep 11:48 < pcjc2> yep 11:48 < Ales> efaaae739bcd0f68197f98fd6b473b469e693773 11:49 < Ales> but it isn't done yet 11:49 < pcjc2> we're looking at different things 11:49 < pcjc2> due to time and bandwidth, I'm not cloning cvs 11:49 < peterbrett> oh, okay 11:49 < peterbrett> fair enough :) 11:49 < pcjc2> I just re-ran a complete log, and md5 summed the output of cvsps 11:49 < Ales> I will be running this again probably tomorrow 11:49 < pcjc2> 0b22874791baef1fb3626e5046725d61 cvsps_patched.output 11:49 < pcjc2> so the patched and unpatched cvsps are producing the same output in this (most simple) case where they start from scratch 11:51 < Ales> does m4 1.4.8 have the multiply bug? 11:52 * peterbrett can't believe that one slipped through the net 11:53 < Dan> ales: I don't think so 11:53 < Dan> I think it was just 1.4.9 11:53 < pcjc2> 1.4.8b might have had it 11:53 < Dan> it is easy to check though 11:54 < Dan> echo "eval(-2/2)" | m4 11:54 < pcjc2> IIRC, it was only divides with -ve numbers, not multiplies which showed it 11:55 < Dan> that sounds right to me too 11:55 < peterbrett> 1.4.8 seems to work for me 11:56 < Dan> dan@bondage 328 % echo "eval(-2/2)" | /export/disk2/tmp/gnum4-1.4.9/bin/m4 11:56 < Dan> 2147483647 11:56 < Dan> dan@bondage 329 % echo "eval(-2/2)" | /export/disk2/tmp/gnum4-1.4.8/bin/m4 11:56 < Dan> -1 11:57 < peterbrett> What did Knuth say about optimisation? 11:57 < Ales> it's the first thing you do 11:58 < Ales> before your write any code 11:58 < dj> programming is the art of debugging an empty file 11:58 -!- jpd [~jpd@66.109.213.29] has joined #geda 12:00 < Dan> hi john 12:00 < Cesar> Carlos: About fopen(file, "wb"): There appears to be an alternative: linking every application against binmode.o when on MinGW. I will investigate the automake magic do that. 12:00 < jpd> Good morning 12:02 < pcjc2> hi John 12:06 < cnieves> Cesar: no problem. I found where it is documented. I'm rebuilding now. 12:06 < cnieves> hi John 12:09 < cnieves> Cesar: patch committed. Please test. 12:13 < Dan> ales: are you happy with the current gnetlist testing framework? 12:14 < Ales> Dan: mostly, except stuart managed to somehow break his spice-sdb tests for make distcheck 12:15 < Dan> what should I look at for an example of adding tests? I need to add one for cascade and pcbpins 12:15 < Dan> how is it broken? 12:15 < Ales> any of them, it's pretty simple 12:15 < Dan> doh! "I have NOT verified 12:15 < Dan> that the known good output is correct." 12:16 < Ales> it doesn't work; I haven't looked at why exactly, probably not copying some dependant file 12:16 < Dan> thats how the tests from the CAD group at work work too ;) 12:16 < Ales> still running git-cvsimport 12:16 < Ales> I sure hope this completes 12:16 < Ales> sometime today 12:16 < pcjc2> what file numbers are you on? 12:17 < pcjc2> I just set one going to see if I could identify where my last incremental tracker lost some of the docs 12:17 < Ales> v1.6 on various symbols 12:17 < Ales> well more than that 12:17 < pcjc2> 1.1.1.1 here ;( 12:18 < pcjc2> its updating a few hundred bytes at a time - how can these take so long!! 12:18 < peterbrett> The symbols take FOREVER :( 12:19 < pcjc2> presumably they haven't changed much since their initial upload though 12:20 < Ales> 42m in .git 12:20 < peterbrett> Ales: It looks like system-gafrc is going to get totally re-written 12:20 < Ales> sounds like fun 12:20 < pcjc2> it will pack that afterwards 12:20 < peterbrett> Ales: after export completed, run "git-repack -a -d", and the size drops like mad 12:20 < Ales> okay, I'll wait till it finishes 12:21 < pcjc2> 6m here - you have a significant head start 12:21 < Ales> and this is only a test run... :) 12:21 < pcjc2> 18m is my tracker which I was actively using 12:21 < pcjc2> (but not bothering to repack too often) 12:22 < cnieves> ivan: https://sourceforge.net/tracker/index.php?func=detail&aid=1730478&group_id=161080&atid=818426 12:22 < Cesar> It's lunchtime. I'll be right back. 12:22 < Ales> yeah, it's going to be food time here too soon 12:22 < Ales> dj is burning the burgers 12:22 < cnieves> ivan: I think it is not up to date. Is it? 12:23 < ivan> I don't know; I thought Ben was going to post a patch 12:23 < ivan> it's a one-line 12:23 < pcjc2> Ales - we probably ought to avoid packing with -a too often 12:23 < Ales> why? 12:23 < pcjc2> means as changes happen, all files get re-packed into a different pack 12:24 < pcjc2> which (I think) users have to get whilst cloning 12:24 < pcjc2> -d means if a new pack makes an old pack redundant, delete the old pack (which is fine) 12:24 < ivan> cnieves: should I just go ahead and send a patch to the tracker? 12:24 < ivan> (I can't test this since I'm not seeing the crash) 12:25 < cnieves> ivan: yes please. However, he is reporting the loop is going backwards, when actually the loop is going forward.... 12:25 < peterbrett> pcjc2: If you're running git-daemon it doesn't matter how often you do or don't repack 12:26 < peterbrett> pcjc2: packs are generated on the fly depending on the client's HEAD. 12:26 < ivan> cnieves: ummm.. I see now 12:26 < ivan> I've moved that code 12:26 < ivan> he refered to the toplevel walking code 12:26 < peterbrett> repacking is only a problem if you're using "dumb" servers such as an plain FTP or HTTP server 12:27 < ivan> the code is now in update_recent_files_menus 12:27 < ivan> i need first to get to the first toplevel so I walk later through all of them 12:28 < ivan> hopefully this will be fixed with Ales' toplevel changes or PeterC's list changes... 12:28 < ivan> in the meantime, i can make a patch for the crash Ben is seeing 12:29 < cnieves> please post it to the tracker, and I'll apply it. 12:29 < ivan> sure, will do in a few minutes 12:29 < ivan> LOL 12:29 < ivan> the check for wid==-1 is already there 12:30 < ivan> so no need to touch that 12:30 < cnieves> Thought so! ;) 12:30 < ivan> my eyes are blurry... :) 12:31 < ivan> perhaps you can then close this bug as solved? 12:33 < cnieves> I wrote a comment "Seems to be fixed in CVS. Please test and report" I'll set the state to pending. If the submitter doesn't update it in 14 days, it will be closed automatically. 12:33 < ivan> cool 12:33 < cnieves> s/I'll set/I set 12:36 < peterbrett> cnieves: That's my usual tactic, works quite well especially for anonymous bugs ;) 12:36 < pcjc2> which file has the wid==-1 ? 12:36 < pcjc2> I'll git blame it 12:37 < ivan> x_menus.c 12:37 < cnieves> The check is already there (x_menus.c) 12:37 < pcjc2> which line? 12:38 < ivan> 436 12:38 < pcjc2> gschem.x_menus.c 12:38 < pcjc2> must have been recent - as I don't see it it my git repo 12:38 < ivan> that's from my patch today :) 12:38 < pcjc2> ah - ok 12:39 < pcjc2> it fixed the bug too 12:39 < pcjc2> good stuff 12:41 < peterbrett> TOPLEVEL.scheme_directory seems to be *only* used for printing a single error message in gschem.c 12:41 < peterbrett> What is more, it's an error message which shouldn't even be printed by gschem, as far as I can tell. 12:41 < ivan> lol 12:42 < pcjc2> Ales - what were the TOPLEVEL changes you had in the pipeline? 12:42 < Ales> but it's used in libgeda in a few places 12:42 < Ales> peterC: I wanted to refactor the rc variables 12:42 < peterbrett> Ales: only in the code for getting and setting it 12:42 < Ales> but I'm not ready 12:43 < peterbrett> Oh wait, it's used in gnetlist for listing available backends 12:43 < bert> Hi :) 12:44 < cnieves> I commited a patch for a bug in the SF tracker. I'll write a comment on the tracker. I'd want the (anonymous) submitter test it again. Should I set the state to pending? 12:44 < peterbrett> cnieves: I don't see why not 12:44 < peterbrett> I suspect it's a fire-and-forget bug anyway 12:44 < cnieves> yup 12:46 < ivan> cnieves: which one? I'm also looking at bug now... 12:46 < Ales> dj has really burned a burger now 12:46 < peterbrett> lol 12:46 < Igor2> :> 12:46 < Ales> smoke detector going off 12:47 < pcjc2> not outside? 12:47 < ivan> :) 12:48 < Ales> no it was raining a lot 12:48 < Ales> its a cookin 12:49 < Ales> 68Meg 12:49 < Dan> I wish we'd have some of that rain 12:49 < Ales> we can ship you a bucket 12:49 < Dan> we got some last night but not enough 12:49 < peterbrett> Hmm 12:49 < Dan> I went fishing last week and the creek had nearly stopped flowing 12:49 < peterbrett> system-gafrc is now 40 lines 12:50 < pcjc2> its glorious sunshine here 12:50 < peterbrett> yeah, it's really lovely :) 12:50 < Dan> its supposed to be 95 F here today 12:50 < Dan> and tomorrow 12:50 < Ales> where are the default symbol paths set? 12:50 < Dan> and in the 90's each day after that for a while 12:51 < Dan> unfortunately "a while" can mean a month 12:51 < peterbrett> Ales: geda-symbols will install "geda-clib.scm" and "geda-font.scm" 12:51 < peterbrett> system-gafrc checks for them and loads them if they exist 12:51 < Ales> what about source libs? 12:52 < Dan> ok, found the problem in my guile pkg that was making libgeda unhappy. 12:53 < Dan> back to building the others 12:53 < peterbrett> Ales: I've ripped most of the doc comments out, as (a) many of them were out of date and (b) I intend to write a "gEDA Guile Reference" on the wiki. 12:53 < peterbrett> Most people seem to specify source libs in project-local gafrc files anyway 12:53 < ivan> peterbrett:: that would be awesome! 12:54 < Ales> yeah, all the source-libraries can be removed 12:54 < ivan> by the way, I really think the site should _be_ the wiki 12:54 < ivan> the wiki now has far more information 12:54 < jpd> Indeed. Guile is daunting, but it's easy once you get past the *gulp* ;-) 12:55 < peterbrett> jpd: I really like the economy of Lisp-like languages :) 12:55 < peterbrett> I've added a "build-path" function to the standard environment too 12:55 < Dan> I don't like porting guile, but I like scheme as a programming language 12:56 -!- sdb [~sdb@ip-66-244-68-18.monroe.lib.in.us] has joined #geda 12:57 < sdb> Hi guys! 12:57 < peterbrett> Stuart! Hi! 12:57 < pcjc2> Hi Stuart 12:57 < ivan> hi 12:57 < cnieves> hi Stuart 12:57 < jpd> Hi Stuart! 12:57 < sdb> Sorry I'm late! I am 1000 miles away from Ales & DJ helping my parents in the 12:57 < sdb> American midwest. 12:57 < dj> hi stuart! 12:57 < jpd> Halfway here. 12:58 < sdb> I have a break from work now, so I'm going to try doing some gEDA related work! 12:58 < dj> I've spent the morning just trying to build pcb on windows *once* 12:58 < Ales> Hi Stuart! 12:58 < pcjc2> oh dear - any specific issues? 12:59 < peterbrett> dj: Why do you put yourself through such torture? 12:59 < Ales> oh git-cvsimport done 12:59 < Ales> now what? 12:59 < Ales> should I repack it? 12:59 < peterbrett> Ales: Can't hurt 12:59 < pcjc2> it did that at the end anyway I think 12:59 < pcjc2> check the file size 12:59 < Ales> 16764 12:59 < peterbrett> That matches my repo pretty closely 13:00 < pcjc2> woo 13:00 < peterbrett> Ales: Now copy the whole .git directory to gaf.git on gpleda.org server in the place gitweb & git-daemon look for repositories 13:00 < Ales> 16764 13:00 < pcjc2> can we see it on gpleda.org now ;) 13:00 < Ales> one second, creating tarball 13:01 < ivan> pcjc2: i think 1705092 should be closed, the warnings are gone 13:01 < pcjc2> one sec 13:02 < pcjc2> oh - those only show up if you compile without gettext 13:02 < ivan> ah 13:02 < ivan> are they still relevant? 13:02 < pcjc2> some dodgy casting with / without "const" 13:02 < pcjc2> I couldn't see a "nice" way round it 13:03 < pcjc2> I did have a stab at fixing them at one point though 13:03 < pcjc2> I'll post you the alternative fixes I came up with 13:03 < ivan> ok 13:04 < pcjc2> will post to dev list 13:04 < ivan> i won't complain -- the trackers are a real pain 13:05 < pcjc2> sent 13:06 < pcjc2> I'm not a fan of bug trackers generally 13:06 < pcjc2> ubuntu's launchpad isn't awful - but its very hard to navigate 13:06 < peterbrett> pcjc2: bug trackers make it far too easy for developers to ignore bugs 13:07 < pcjc2> it would be nice to get a summary email of bugs assigned to each person, say every fortnight / month 13:07 < pcjc2> with all bugs being assigned to "someone" by default 13:07 < sdb> Hmmmm, I have a dumb question. Perhaps I missed something this morning. 13:07 < ivan> i think i will focus some of my dev time to bug solving... 13:07 < sdb> With all hte talk about git, has the CVS sererver been shut down? 13:07 < sdb> I'm getting this error when trying to refresh: 13:07 < pcjc2> GTK have this problem, and they adopted a similar solution, someone goes and emails a summary of patches which need reviewing / applying etc.. 13:07 < sdb> connect to cvs.seul.org(128.31.0.34):2401 failed: Connection timed out 13:08 < sdb> Or am I blocked to the CVS port here? 13:08 < Ales> no cvs should be fine 13:08 < sdb> Ales, thanks. 13:08 < peterbrett> Meh crappy ISPs :( 13:08 < Ales> peter(s): git.gpleda.org 13:10 < pcjc2> Carlos: you added a patch which logs a warning for missing pixmap 13:10 < pcjc2> my win32 patch needed something (not NULL) returned IIRC 13:10 < pcjc2> so I added: + wpixmap = gtk_image_new_from_stock(GTK_STOCK_MISSING_IMAGE , GTK_ICON_SIZ 13:10 < pcjc2> E_SMALL_TOOLBAR); 13:10 < pcjc2> seems like a nice addition to the logged warning message 13:11 < peterbrett> Ales: touch gaf.git/git-daemon-export-ok 13:11 < Ales> oh yeah sorry forgot 13:11 < pcjc2> as a nicety, you could edit .git/description 13:11 < peterbrett> Also I think you need to edit gaf.git/description 13:11 < peterbrett> lol 13:12 < Ales> in stero 13:12 < Ales> stereo 13:12 < ivan> alright folks, i have to go now. it's been an interesting sprint, keep up the good work! bye 13:12 < peterbrett> ivan: Have a good one. 13:12 < pcjc2> bye! 13:12 -!- ivan [~ivan@sdgate2.cvjetno.sczg.hr] has quit [Quit: Leaving] 13:12 < pcjc2> We should try and setup the graphic log stuff which repo.or.cz uses at some future time, its quite nice 13:12 < Ales> done done 13:13 < pcjc2> its all very nice ;) 13:13 < Ales> I like the changeset feature already. :) 13:14 < pcjc2> and you can get arbitrary snapshots 13:14 < Ales> soo nice. 13:14 < Ales> this is going to be so much easier 13:14 < pcjc2> downloaded as a tar ball at any point in the history 13:14 < peterbrett> Ales: Have you tried gitk? 13:14 < Ales> stuart: any progress on the cvs problem? 13:14 < bert> Hi Stuart :) 13:14 < pcjc2> heck - I often prefer gitweb to the raw command line tools in that respect! 13:14 < Ales> just finished a git-clone 13:15 < peterbrett> Ales, sdb: Sounds like a good time to try git-cvsserver ;) 13:15 < Ales> I need to figure out how to install that. 13:15 < sdb> Ales, no. I think my port is blocked. I'm trying this at the public library. 13:15 < Ales> really? 13:15 < Ales> (15) 13:14:26 [laptop:git] $ 13:15 < sdb> I am thinking about trying out git. What port does it use? 13:15 < Ales> ack 13:15 < Ales> telnet cvs.seul.org 2401 13:15 < Ales> try that 13:15 < pcjc2> can use the web port if you want it to 13:16 < peterbrett> Ales: In future when importing patches (like you did in 98c5aa9857df54091624160a6505dd072ff9e3b3) you should set the GIT_AUTHOR_NAME and GIT_AUTHOR_EMAIL environment variables, and import them one at a time rather than all together 13:16 < peterbrett> Ales: As you can see from the git log, it records author and committer separately 13:16 < Ales> I haven't done any git imports yet 13:16 < peterbrett> Ales: I know, just for future reference ;) 13:16 < Ales> are you saying I should have set those first? 13:16 < Ales> oh yeah.. I'll add that to my list. 13:17 < sdb> Ales, "telnet cvs.seul.org 2401" seems to time out too. 13:17 < Ales> we should probably get a geda-git development wiki page 13:17 < peterbrett> indeed 13:18 < Ales> I'll start one once I figure out how to use git. :) 13:19 < bert> Stuart: you mentioned doxygen w.r.t.pcb in the february sprint, are you serious about doxygen generated docs for pcb ? 13:20 < sdb> Bert, I actually did put some doxygen comments into 13:20 < sdb> some of hte PCB source. I just haven't submitted the patches. 13:21 < Ales> gitk is slick too 13:21 < Ales> all offline too, way cool 13:21 < sdb> The reason is partially laziness, and partially because PCB alreayd 13:21 < bert> ah, I see, I was woundering as I didn't see any. 13:21 < sdb> has its own, home-grown method to generate documentation from comments. 13:21 < sdb> The machine with comments is back in BOston, so I wont do that this time. 13:22 < Joely> ah, sorry this is kinda off topic and a silly question, but this is my first _real_ fpga project...i'm trying to make a customizeable processor (ranging from 1 core -> 8) and the timer bit width is a constant dependent on what? i'm looking at the source code to Simply Risc's implementation of Sun's T1, and it just sets it as a constant of 16 bits. i'm wondering why that is, if it needs to be scalable per core. 13:22 < peterbrett> Joely: Sorry, what was the question? 13:22 < Ales> we are demoing the design flow and programs to Larry the usability expert 13:23 < peterbrett> *blink* 13:23 < Joely> what is the timer bit width dependent on as it is set separately in another file called `s1_defs' 13:23 < peterbrett> Joely: Are you using iverilog? 13:23 < Ales> john: got your e-mail, I'll check it in 13:24 < Joely> peterbrett, yes, for simulation 13:24 < jpd> Thanks, Ales. 13:24 < peterbrett> Joely: Hmm. I really think you'd be better off sending an e-mail to the geda-user mailing list 13:24 < bert> sdb: are you going to push this in the future or is it just for educational purposes ? 13:24 < peterbrett> I don't think there are any iverilog specialists around today 13:25 < cnieves> pcjc2: didn't know there was a missing image pixmap. That's fine. 13:27 < sdb> Bert, ummmm, push what? I'm sorry I'm confused. 13:27 < sdb> Bert, do you mean DOxygen docs for PCB? 13:27 < bert> sdb: implementing doxygen comments or stick to the home-grown stuff ? 13:28 < sdb> Bert, that decision is not up to me, but rather is up to DJ, Dan, and Harry. 13:28 < pcjc2> Ales: git-cvsimport has a way of mapping usernames like "pcjc2" to the full "Peter Clifton" "pcjc2@cam.ac.uk" as it imports 13:28 < sdb> When we last discussed it, I think the PCB developers were sorta lukewarm about 13:29 < sdb> doxygenating their code, but they weren't opposed as long as the mark-up was 13:29 < pcjc2> but we'd need to produce a list of all users. The alternative would be to leave it as is, which is also workable 13:29 < sdb> compatible iwht their home-grown stuff. I think that makes perfect sense 13:29 < dj> I'm ok with doxygen, as long as we can generate pcb docs with it, like we do today. 13:29 < sdb> I just haven't done anything about it since... .well, I have a number of projects running 13:29 < sdb> right now and DOxygenating PCB fell down the priority list. 13:30 < bert> sdb: OK, I understand. 13:30 < sdb> THe reason I wanted it was so that I could get a handle on how the code worked. 13:30 < bert> sdb: It would be my first choice either, it's rather cumbersome 13:30 < peterbrett> pcjc2, Ales: I vote for leaving it how it is. 13:30 < Ales> Peter: is this related to the environment variables mentioned before? 13:30 < bert> s/would/wouldn't/ 13:31 < pcjc2> yes, for example - what I use: 13:31 < pcjc2> cat ~/.gitconfig 13:31 < pcjc2> [user] 13:31 < pcjc2> name = Peter Clifton 13:31 < pcjc2> email = pcjc2@cam.ac.uk 13:31 < pcjc2> so when I make a commit, those details are recorded 13:31 < pcjc2> git-cvsimport can convert the CVS user names to be like those too 13:31 < bert> sdb: and it doesn't bring new features for the pcb-users :) 13:32 < peterbrett> heh. Apparently the conflicting filenames in the default symbol libraries will be fixed "soon" (tm). 13:33 < pcjc2> Bert - thanks for posting those images 13:33 < sdb> Bert: That's true. For me it was a mechanism to understand the PCB code, with the 13:33 < sdb> vision that I might fiddle around with some of my usability peeves after 13:34 < sdb> I understood it. That project is dead in the water right now. :-( 13:34 < pcjc2> "soon" (tm) I'll have to get around to refreshing my page navigation patches - I've already had a stab at recovering some of the GedaList based stuff 13:35 < bert> sdb: for me it would serve the same purpose: go through the code base and get a grasp what is going on :) 13:36 < peterbrett> al: which of the many gnucap sites is the "current" one? 13:37 < bert> sdb, dj: if I understand DJ's response well it would be implenetd in between releases so their would be no pieces of comments dangling outside the docs 13:37 < bert> s/implenetd/implemented/ 13:38 < bert> I should find a better keyboard instead of this typo generator :) 13:38 < sdb> In my vision, we would just add comments to the functions slowly, over time. 13:38 < sdb> There is no need (to users, anyway) for every fcn to be documented. 13:38 < pcjc2> Anyone fancy taking a glance at some patches at http://repo.or.cz/w/geda-gaf/pcjc2.git ? 13:38 < sdb> It's really just an exercise to help developers better understand hte code. 13:38 < pcjc2> I'm opening up the floor with RFC to adding GedaList (A gobject) to libgeda 13:39 < pcjc2> The selection code changes ought to be fairly complete with it 13:39 < pcjc2> but the very last patch is a mess - ignore that (other than to see what it might look like eventually) 13:40 < pcjc2> Also, the first patch makes a libgeda/include/geda.h where I've been putting new stuff under development - it keeps the conflicts down 13:40 < peterbrett> pcjc2: No major objections here 13:40 < peterbrett> pcjc2: libgeda already links against gobject anyway 13:40 < pcjc2> it already links against GTK - that doesn't make it an ideal situation ;) 13:40 < peterbrett> pcjc2: true ;) 13:40 < dj> The docs we currently have are primarily for the API references, and keeping the "docs" in the source makes it a lot easier to keep them in sync. 13:41 < pcjc2> anyway - I am of the opinion that gobject brings us benefits, and that we ought to use them 13:41 < dj> The sources are parsed to build parts of pcb.texi, which is processed into everything else. So it's all at build time that the differences between what we have and what doxygen wants matter. 13:41 < bert> sdb, dj: so to you guys it would be OK if I put a patch on SF every now and then 13:41 < sdb> pcjc2: What version of GTK introduced the gobject stuff you want to use? 13:41 < dj> Sure, or on the -dev mailing list. 13:41 < peterbrett> Ales: Any decisions on whether you want to keep the import you've got, or make a new one? 13:41 < peterbrett> sdb: 2.0 13:42 < dj> One could certainly *add* doxygen comments, without changing the existing API doc-comments, too, and migrate the rest later. 13:42 < pcjc2> its GLib stuff. Ideally, libgeda shouldn't depend against GTK 13:42 < Ales> I'm going to reimport after the sprint is done 13:42 < sdb> Bert: Go for it! I am not a PCB develper, so DJ's opinion is more valuable than mine here. 13:42 < peterbrett> Ales: Are you going to use a usermap file to put people's proper names/email addresses into the import? 13:43 < Ales> docs on doing that? 13:43 < Ales> is it worth doing? 13:43 < sdb> Peterbrett: OK. 13:43 < pcjc2> man git-cvsimport 13:43 < peterbrett> Ales: It makes it look nice ;) 13:43 < pcjc2> under "-A" 13:44 < pcjc2> I'm not sure - does it set the commit name and email as well as the author's ? 13:44 < peterbrett> pcjc2: I believe it sets both to the same thing 13:44 < bert> sdb, dj: it's a go then :) I think I will start with HID's 13:44 < Ales> okay, I'll do that for all the contributers. 13:45 < peterbrett> I recommend also adding a robots.txt which stops Google from indexing the gitweb 13:45 < pcjc2> PeterB: For the sake of our email addresses, or the repo's content not being useful to google search? 13:45 < peterbrett> both 13:45 < sdb> OK, now my next question. I just installed git on my traveling laptop. 13:46 < sdb> Could some kind soul tell me the magic incantation (command + git repo) 13:46 < pcjc2> I already get private emails relating to gEDA bugs occasionally - don't want lots more ;) 13:46 < sdb> I need to check ou the latest stuff? 13:46 < Ales> I've never gotten those 13:46 < peterbrett> sdb: hopefully you have cogito 13:46 < peterbrett> sdb: cg clone git://git.gpleda.org/ 13:46 < Ales> note, the current git repository is temporary 13:46 < sdb> PeterBrett: Nope. I'll install that now..... 13:47 < peterbrett> sdb: for testing only ;) 13:47 < pcjc2> with any luck, when its switched over, the import will be the same, and it will all work fine - just with more recent commits in it 13:47 < Ales> yup 13:47 < sdb> PeterBrett: My question is: is that repo in sync wiht the master CVS repo? 13:47 < pcjc2> ok - (changes his mind) the SHA1 IDs of the commits will change when we import with the pcjc2 -> Peter Clifton type swiching 13:47 < peterbrett> sdb: At the moment, but I'd hold off until the "official" switchover in a day or so 13:48 < peterbrett> pcjc2: correct 13:48 < Ales> shouldn't it be: cg clone git://git.gpleda.org/git/gaf 13:48 < Ales> I don't know if I like that extra "git" in there 13:48 < peterbrett> cg clone git://git.gpleda.org/gaf.git ?? 13:49 < Ales> nope. 13:49 < peterbrett> Um 13:49 < peterbrett> I can't clone :-( 13:49 < Ales> cg clone git://git.gpleda.org/git/gaf works for me 13:49 < peterbrett> cg clone git://git.gpleda.org/git/gaf.git works for me 13:50 < sdb> Another issue: I don't want to trash my current version of geda/gaf yet. So if I do 13:50 < peterbrett> Maybe you could put that in gaf.git/description ? 13:50 < Ales> what is "that"? 13:50 < sdb> cg clone git://git.gpleda.org/git/gaf will that put it in a separate "git" directory (which is what I want.) 13:51 < peterbrett> It'll create a gaf directory 13:51 < peterbrett> 'cg clone git://git.gpleda.org/git/gaf git' will create a git directory 13:51 < sdb> PeterBrett: Thanks! 13:51 < Ales> are we okay with the path "git://git.gpleda.org/git/gaf" ? 13:52 < Ales> I guess I can change it 13:52 < peterbrett> Ales: git://git.gpleda/gaf[.git] would be more "canonical" 13:52 < sdb> Hmmmm... which port does git use? It seems to be taking a long time to clone something..... 13:52 < peterbrett> sdb: ... no idea, sorry :-/ 13:53 < sdb> Uh oh -- "fatal: unable to connect a socket (Connection timed out)" 13:55 < bert> I just cloned gaf on my FC5 box: no problems whatsoever :) 13:55 < Ales> bert: using git or cg ? 13:55 < peterbrett> I just cloned it using cg, took a while but that's normal :) 13:55 < Ales> hmmm 13:55 < Ales> I got an error before, but it completed 13:56 -!- sdb [~sdb@ip-66-244-68-18.monroe.lib.in.us] has quit [Read error: Connection reset by peer] 13:56 < Ales> remote: Total 19, written 19 (delta 0), reused 19 (delta 0) 13:56 < Ales> Indexing 19 objects. 100% (19/19) done 13:56 < Ales> fatal: pack: not a valid SHA1 13:56 < Ales> New branch: ef17dab2d16bbc8a0a325501b8dfac6c8d440d85 13:56 < bert> cg clone git://git.gpleda.org/git/gaf 13:56 < peterbrett> Ales: That's odd. 13:56 < bert> sdb: did you git-init first ? 13:56 < peterbrett> bert: You shouldn't need to 13:56 < Ales> I'll try it again 13:57 < bert> Well, It did no harm I guess ? 13:57 < Ales> okay I'm going to change the path to gpleda.org/gaf 13:57 < peterbrett> Ales: Can you put the clone URL in the description? 13:57 < Ales> sure 13:57 < peterbrett> If you put it in a second paragraph it won't clog up the projects list 13:57 < Ales> oh okay 13:58 -!- sdb [~sdb@ip-66-244-68-18.monroe.lib.in.us] has joined #geda 13:59 < Ales> how can I see that second paragraph? 13:59 < bert> du gives 46M disk usage 13:59 < bert> took me about 60 seconds 13:59 < peterbrett> bert: That's repo (16MB) + checked-out sources (30MB) 14:00 < pcjc2> Ales: The SHA1 between our trackers match up until about 2003 14:00 < bert> Well df gives 58GB free so no problem yet :) 14:00 < pcjc2> http://git.gpleda.org/?p=gaf.git;a=commit;h=24715da4504cbef6dd110451bfc4a25b426d1781 14:00 < pcjc2> that commit 14:01 < pcjc2> I'd not be too surprised though - as it could be a file permission different or something 14:01 < pcjc2> perhaps there was CVS black magic conducted at some point which touched one of those files 14:01 < Ales> yeah 14:02 < pcjc2> it should be possible to rebase any existing git repos 14:02 < pcjc2> it might take a bit of "force", as git won't think the trees match - but it can be done 14:02 < pcjc2> either with git-rebase, or my new preferred method (seems safer), exporting a series of patches, then re-applying on a different branch 14:02 < peterbrett> pcjc2: The repo.or.cz repo will become a mirror of the official one 14:02 < Ales> I put the url in the description 14:02 < Ales> as the second paragraph 14:03 < peterbrett> It really doesn't seem to be showing up, does it? 14:03 < peterbrett> Hmm. 14:03 < pcjc2> Indeed - although I've not been using the repo.or.cz tracker, I kept my own in sync since yours lost a noscreen patch or so - and have been using it since 14:03 < Ales> yeah 14:03 < Ales> we can put that sort of info into the geda git wiki 14:03 < pcjc2> once you have it up to a point, its low bandwidth and effort to keep it updated anyway 14:03 < peterbrett> never mind. *shrug* 14:04 < pcjc2> seems mine dropped some docs recently 14:04 < bert> after coning I thought of running gitk (and qgit) and they both barf on something and bail out, is this normal ? 14:04 < pcjc2> I don't think the tracker likes you updating whilst commits are going on! 14:04 < bert> s/coning/cloning/ 14:04 < pcjc2> Bret: are you in the dir with .git in it? 14:04 < bert> I has gaf and .git dirs 14:05 < Ales> I'm going change the git repository 14:05 < Ales> ^repository^UL 14:05 < pcjc2> run gitk from the checkout's toplevel dir 14:05 < pcjc2> that is the one with the .git dir in it 14:06 < pcjc2> if it bails, post the error message 14:06 < bert> I made a git.gpleda.org dir and that's where I'm in right now 14:06 < Joely> peterbrett, how's this?? hehe http://archives.seul.org/geda/user/Jun-2007/msg00092.html 14:06 < sdb> Another git question: doing a clone git://git.gpleda.org/geda/gaf geda-git" seems to time out. 14:06 < Ales> possible you isp is blocking you? 14:06 < sdb> Therefore, I tried usign http: "git clone http://git.gpleda.org/geda/gaf geda-git" 14:06 < Ales> did that work? 14:07 < sdb> This gives me the following error: Cannot get remote repository information. Perhaps git-update-server-info needs to be run there? 14:07 < sdb> Ales, yeah, I think I am experienceing port blockage. 14:07 < Ales> okay, I don't think I've setup the http part of things 14:07 < sdb> Ales, ummmm, is it easy to do? If so, setting it up would be great! 14:07 < pcjc2> Ales - is the repo gaf.git dir directly accessible by web? 14:08 < pcjc2> If those dirs are somewhere the webserver can get at them, to clone - all you need is that URL 14:08 < sdb> Indeed, that would be an added benefit of git -- we could grab stuff via http, which uses a port almost nobody blocks. 14:08 < bert> [bert@steiger git.gpleda.org]$ gitk & 14:08 < bert> [1] 11776 14:08 < bert> [bert@steiger git.gpleda.org]$ Error in startup script: child process exited abnormally 14:08 < bert> while executing 14:08 < bert> "close $refd" 14:08 < bert> (procedure "readrefs" line 47) invoked from within"readrefs" (file "/usr/bin/gitk" line 6325)[1]+ Exit 1 gitk 14:08 < Ales> I need to setup an alias for that 14:08 < pcjc2> ok - I got that before on a bugged checkout 14:09 < pcjc2> urm - I'm not sure.. let me think for a sec 14:09 < peterbrett> sdb: It's not particularly a nice way to do it. You'll get *much* larger downloads, and it stops Ales from running git-repack on the server to save disk space 14:10 < pcjc2> Bert: try using "git-clone" for me 14:10 < pcjc2> git-clone URL 14:10 < al> The main gnucap site is gnucap.org. Other gnucap.* (like gnucap.net) point to the same IP as gnucap.org. The gnu site (www.gnu.org/software/gnucap) is from CVS. The one on geda.seul.org is in need of updating. 14:11 < pcjc2> I have to go now I think 14:11 < pcjc2> has been a good productive sprint so far, I look forward to reading the rest of the logs later 14:12 < Ales> I'm going to keep banging on git 14:12 < Ales> git clone http://git.gpleda.org/git/gaf.git 14:12 < Ales> works for me 14:12 < Ales> so what's this about not being able to run repack if we allow this? 14:13 < pcjc2> not can't, just ought not repack into a single pack file 14:13 < pcjc2> if you allow http access, when people clone, they tend to grab large packs of the server 14:13 < pcjc2> and if you repack, and they clone / fetch again, they have a big download to do 14:13 < Ales> okay, I might disable this since I do not have infinite bandwidth 14:13 < Ales> okay 14:14 < peterbrett> Ales: New system-gafrc / symbols Scheme files are working nicely 14:14 < sdb> Ales, you rock! That incantation seemed to work! 14:14 < bert> pcjc2: same result it gives: 14:14 < pcjc2> you could let repo.or.cz mirror it 14:14 < bert> [bert@steiger git.gpleda.org]$ gitk & 14:14 < bert> [1] 11922 14:14 < bert> [bert@steiger git.gpleda.org]$ Error in startup script: child process exited abnormally 14:14 < bert> while executing 14:14 < bert> "close $refd" 14:14 < bert> (procedure "readrefs" line 47) 14:14 < bert> invoked from within 14:14 < pcjc2> and point people to the http url f that 14:14 < Ales> sdb: don't get too attached to it. 14:14 < bert> "readrefs" 14:14 < bert> (file "/usr/bin/gitk" line 6325) 14:14 < bert> [1]+ Exit 1 gitk 14:15 < sdb> Ales: :-) I just need it this once since all my ports are blocked here. 14:15 < sdb> Ummm, one question, though: Is the stuff I'm downloading the latest & greatest? 14:15 < Ales> sdb: nope 14:15 < sdb> Or is the stuff in regular CVS newer? 14:15 < Ales> it's a test only 14:15 < sdb> Oh, OK. 14:15 < Ales> the stuff in CVS is way newer 14:15 < bert> du now gives 40M and it took almost 2 minuts 14:16 < Ales> okay git-cvsserver is next 14:16 < pcjc2> bye! - I'll leave my IRC open to make my own log 14:16 < Ales> see ya peter1 14:16 < sdb> Bye! 14:16 * bert is away: I'm busy doing something else 14:17 < peterbrett> I've persuaded my lovely girlfriend to make dinner for me, so I'm still around for a bit :) 14:17 < Ales> blah 14:17 < Ales> git-cvsserver that is. :) 14:17 < peterbrett> misbehaving? 14:18 < Ales> no, I'm just reading the install docs 14:18 < Ales> though I better change the repository URL first 14:20 < Ales> I'm bringing down git-daemon 14:20 < Ales> complain now 14:21 < Ales> git-clone git://git.gpleda.org/gaf 14:21 < Ales> much better 14:22 < peterbrett> wonderful 14:22 < Ales> the url has changed to: git://git.gpleda.org/gaf 14:24 < Ales> where the heck is git-cvsserver 14:25 < peterbrett> Install the git-cvs package 14:26 < Ales> yeah that would help 14:26 < Ales> :) 14:26 < Ales> ah, there it is 14:26 < Dan> john: glad that the aliasing code was easy to put in 14:26 < Dan> that was the intent of the design 14:28 < jpd> Yes. Thanks. 14:29 < Dan> I'll bet other backends need it 14:29 < Dan> my guess is some need it because they munge the names unsafely and I'll bet some need it because the geda name space is larger than the target format 14:29 < Ales> speaking of which, I just commited that backend fix from John 14:29 < Ales> thanks 14:30 < jpd> Maybe. Right now I'm working on a Mathematica back end for symbolic circuit analysis. 14:31 < sdb> John, that's a cool project! What kind of symbolic analysis do you plan to support? 14:32 < jpd> Initially, symbolic transfer functions and impedances as functions of complex frequency. The usual linear network theory stuff. 14:34 < sdb> JPD: So you need to have a program which takes a netlist and turns it into a transfer fcn. 14:34 < sdb> How does that work? 14:35 < sdb> (I mean, SPICE can do that, but getting from a netlist to a transfer fcn is not trivial.) 14:36 < jpd> I'm writing a gnetlist back end that generates the KCL equations and equates pin voltages to node voltages. 14:36 < jpd> Also generates invocations of the model functions (still designing that). 14:37 < jpd> Then I'll have a set of Mathematica functions to put it all together, solve it algebraically. 14:38 < jpd> SPICE can't do *symbolic* transfer functions. 14:39 < sdb> Yeah, SPICE doesn't do symbolic stuff, which is why your ideas sound very interesting. 14:39 < sdb> OTOH, your stuff assumes linear circuits only, I suppose. 14:39 < peterbrett> Wow, they've speeded up `install' an awful lot in Fedora 7! 14:40 < sdb> OK, now that I have checked something out using git, I am ready to work. 14:40 < sdb> Carlos: You filed a bug 1705025 saying that gattrib doesn't work with un-numbered 14:40 < jpd> In principle it can do anything SPICE can do with the right models. Can feed the equations to a numerical diffeq solver rather than to an algebraic solver. 14:41 < jpd> I see no point in competing with Al, though ;-) 14:41 < sdb> refdeses. That's true. I can't imagine how to deal with attaching attributes to unnumbered 14:41 < sdb> refdeses. Carlos, do you have any thought on how to identify an un-numbered part for 14:41 < sdb> attribute attachment? 14:42 < cnieves> sdb: I filed a bug because gattrib crashed.... 14:42 < cnieves> Now I think about it... if you know the object pointer each row entry belongs to, it shouldn't be so difficult. can you know it? 14:43 < sdb> Carlos: Oh. OK. I am now downloading your test cases. 14:43 < sdb> Carlos, as for the object pointer, yes, the software can get the component 14:43 < sdb> via the object pointer, but that has no meaning to the poor user. 14:44 < sdb> JPD: Yeah, it can do anything SPICE can do once you have the right solvers and models. 14:44 < sdb> By that time it will also contain several hundred thousands of lines of code, like SPICE! :-) 14:45 < cnieves> sdb: no meaning, but then you are being coherent and gattrib won't crash 14:47 < Ales> ack! missing dependancy for git-cvsserver that wants to bring in apache 1.x 14:47 < Ales> not going to fly 14:47 < peterbrett> Ales: Install git-cvsserver from source? 14:48 < Ales> Can't locate DBI.pm in @INC (@INC contains: /usr/share/perl/5.8.8 /etc/perl /usr/local/lib/perl/5.8.8 /usr/local/share/perl/5.8.8 /usr/lib/perl5 /usr/share/perl5 /usr/lib/perl/5.8 /usr/share/perl/5.8 /usr/local/lib/site_perl .) at /usr/bin/git-cvsserver line 2062. 14:48 < peterbrett> That's pathetic 14:48 < Ales> need to make sure DBI.pm really lives in libapache-dbi-perl 14:48 < peterbrett> You could install DBI.pm directly from CPAN 14:49 < sdb> Carlos, yeah I get a segfault following your bug's instructions. Indeed, the segfault 14:49 < sdb> happens to me *before* I finish your instructions. 14:49 < Ales> blah, maybe I won't worry about git-cvsserver for a little bit of time 14:50 < sdb> Hmmmmm..... I know that gattrib doesn't like to load one design on top of another one. 14:50 < peterbrett> sdb: Actually, I have found that gattrib has a number of "lose all your work"-style bugs. 14:51 < peterbrett> To be fair, you have squished all of the ones I've got round to reporting 14:51 < sdb> Peterbrett: Please report the ones whihc you haven't reported yet. I won't promise to 14:51 < sdb> fix them today (or tomorrow), but at least I can then get to them at some point. 14:51 < peterbrett> sure, when I have some dead time (or I run across one again) 14:52 < peterbrett> so... 14:52 * peterbrett looks at the next thing on the to-do list 14:53 < dj> I'm still here folks, just giving our guest Larry a tour of the gEDA workflow. 14:53 < sdb> DJ, does he have any opinions you'd like to share? 14:53 < dj> I have a short list but he seems to think we've got it pretty streamlined 14:56 < dj> ales has been keeping a list too. 14:57 < sdb> Dan, I am looking at a gattrib bug "[ 1723885 ] gattrib crashes on file save". 14:57 < sdb> The reporter is using FreeBSD. I run this on my FC2 box, and 14:57 < sdb> I don't get a crash. 14:57 < sdb> Could you please take a look at his bug and see if you can reproduce it? 14:57 < sdb> I am wondering if it is a BSD thing, and I am doing somethign non-portable. 15:02 < sdb> Ales, yesterday you asked me which version of guile I was using when I reported 15:02 < Ales> yeah? 15:02 < sdb> that hitting "g" caused gshem to segfault. I am using 1.8.1 15:02 < Ales> okay 15:02 < Ales> yeah, I think we know what was causing that 15:02 < Ales> speaking of which, has somebody checked in the fix? 15:03 < cnieves> yup 15:03 < Ales> great 15:03 < Ales> I haven't read through the commit logs yet 15:03 < peterbrett> Ales: system-gschemrc will still need to be processed with configure 15:04 < peterbrett> Because of the gschem-version thing. 15:04 < Ales> ah yes 15:04 < peterbrett> Is it actually used/useful 15:05 < Ales> it is used to keep gschem and rc versions to match 15:05 < Ales> that used to be a problem with people trying to use old rc versions with newer software 15:05 < Ales> and vice versa 15:12 < peterbrett> I don't think the gschem sort-component-library option does anything any more 15:12 < peterbrett> The dialog just shows the component sources in the order returned by s_clib_get_sources(), which is alpha case-insensitive 15:13 -!- dj is now known as dj-mon 15:14 -!- dj [~dj@duopoly.delorie.com] has joined #geda 15:15 < dj> pcb ideas from larry... 15:15 < dj> locked elements on a board should be visibly identifiable 15:15 < dj> or, the cursor should add a padlock icon when it's over a locked item 15:16 < dj> when rotating an element, the text's orientation should remain constant even as the text moves (i.e. text moves, but not rotates) 15:16 < dj> bigger grid dots 15:16 < dj> the ability to crob the board by selecting a rectangle on the screen, rather than using the board size dialog. 15:16 < peterbrett> +1 15:16 < dj> which? 15:17 < sdb> DJ: one of my peeves about PCB is that when you hit the del key, it doesn't delete 15:17 < peterbrett> last 15:17 < sdb> the selected items. Rather, it deletes the object under the cursor. 15:17 < sdb> What does Larry think about that? 15:17 < peterbrett> Yes, :DeleteSelected() is my most common command 15:17 < dj> IMHO delete should "delete selected" if there's something selected, else delete what's under the cursor. 15:17 < al> jpd: Have you looked at Mathomatic? 15:18 < dj> A lot of tools should auto-detect selected vs none selected 15:20 < dj> I think once we have real layer types, editing the board outline will become more important and flexible, too. 15:21 < sdb> As for my delete bleat, my beef is that PCB often wants to do 15:21 < dj> Should pcb be able to zoom out further than board extents? It might confuse the scroll bars. 15:21 < sdb> del del del del 15:21 < sdb> when I want to do