Back
[11:47:20] <mozmck> seb_kuzminsky: you around? I'm thinking that libgnomeprintui2.2-dev should be removed as a build dependency for systems newer than wheezy and precise, because it is no longer available for these systems.
[11:47:36] <mozmck> Looks like the only thing it affects is printing from ClassicLadder.
[11:52:30] <mozmck> could it be moved from the control file to the configure script as an EXTRA_BUILD dependency for the systems that support it?
[11:59:36] <seb_kuzminsky> mozmck: i agree
[12:00:07] <seb_kuzminsky> i had started to do that work, err, last fall, but i got sidetracked and dropped it by mistake
[12:00:21] <seb_kuzminsky> now where did i put that branch....
[12:00:26] <mozmck> Looks like it is only in precise and lucid on ubuntu
[12:01:40] <mozmck> Oh, another thing - how can I make debian/configure look for LinuxMint 17* without having a separate entry for each one? 17, 17.1 and etc are all based on Ubuntu Trusty
[12:01:41] <seb_kuzminsky> yep, trusty and newer dont have it, and jessie and newer dont have it
[12:01:58] <seb_kuzminsky> yeah, that's an annoying thing about our debian/configure, isn't it?
[12:02:23] <seb_kuzminsky> when i went to fix the libgnomeprintui thing i got fed up with that and rewrote our debian/configure to be less wonky
[12:02:34] <mozmck> Is there a way to make the statements drop through like they can in C?
[12:02:53] <mozmck> fix it by making classicladder not depend on it?
[12:03:19] <seb_kuzminsky> ah, here's the branch:
http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=shortlog;h=refs/heads/seb/master/new-deb-configure
[12:03:46] <seb_kuzminsky> i'd love it if you finished that branch (i think it's good except for being out of date, and doing the tcl dependency wrong)
[12:04:04] <seb_kuzminsky> i'd also be happy if you did the quicker, easier, safer thing of just fixing (one more time) the old debian/configure we currently have
[12:04:05] <mozmck> by statements drop through I mean like a switch statement where you can stack case statements without break;
[12:04:32] <mozmck> I'll look at that branch and see what I can figure out
[12:05:07] <seb_kuzminsky> thanks!
[12:07:28] <mozmck> looks like you changed configure a good bit!
[12:07:36] <seb_kuzminsky> yeah, it's a total rewrite
[12:07:46] <seb_kuzminsky> to hopefully be cleaner & easier to maintain
[12:08:28] <seb_kuzminsky> i originally did that work for the ubc branch, before it became clear that that was a fork, not a serious merge candidate... so then i backported that debian/configure to work with our code
[12:08:44] <seb_kuzminsky> oh, i see now that i already did the gnomeprintui thing in that version of debian/configure
[12:09:12] <mozmck> It does look cleaner and easier to extend.
[12:09:20] <seb_kuzminsky> http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=blob;f=debian/configure;h=9ab5a37a1b9388580d074f0e336950537217e29a;hb=bca5f62998fa83a418bd2c663aaa53a214e1af03#l220
[12:10:16] <seb_kuzminsky> so it just needs to be rebased onto a more up-to-date place in our history (probably the tip of 2.7)
[12:10:32] <seb_kuzminsky> at which point it will break the tcl test that jepler added to show me the error of my ways
[12:10:50] <seb_kuzminsky> and then we'll fix that (by specifying explicitly the tcl version again) and all will be hunky-dory
[12:11:00] <mozmck> oh, for the tcl version?
[12:11:14] <seb_kuzminsky> hm, i guess it's in better shape than my memory told me...
[12:11:47] <mozmck> heh, you might can finish it faster than I could at this point then.
[12:11:52] <seb_kuzminsky> yeah, i had mistakenly thought tcl versions were more interchangeable than they really are, and that it'd be ok to just use the default "tcl" package instead of the "tclX.Y" package (and same for tk)
[12:12:41] <mozmck> default tcl/tk version on 14.04 is 8.6 now
[12:12:57] <seb_kuzminsky> yeah
[12:13:08] <seb_kuzminsky> oh yeah, that reminds me, axis breaks on tcl8.6
[12:13:16] <seb_kuzminsky> i think it triggers a bug in tcl, amazingly
[12:13:25] <mozmck> oh? I think I've been running with that...
[12:13:35] <seb_kuzminsky> https://sourceforge.net/p/emc/bugs/411/
[12:13:45] <seb_kuzminsky> which version of tcl.deb do you have?
[12:13:49] <seb_kuzminsky> maybe upstream has fixed it
[12:15:38] <mozmck> Looks like I'm using tcl8.6-dev
[12:16:01] <mozmck> 8.5 is installed as well, but not the 8.5-dev
[12:17:15] <seb_kuzminsky> which version of tcl8.6-dev? (dpkg -s tcl8.6-dev | grep Version)
[12:18:05] <seb_kuzminsky> i'd love to have the libgnomeprintui thing fixed so we can support trusty & jessie, but it's not at the top of my personal to-do list
[12:18:07] <mozmck> oh, I just found a neat way to install dependencies before you have the packages built. you have to have devscripts and equivs installed
[12:18:54] <seb_kuzminsky> i want to deal with rob's arc-radius/tp-fixes branch, arceye's ini updates, the new standalone-motion testing project, and the new po4a documentation translation project first...
[12:19:08] <mozmck> then in the debian directory run mk-build-deps -B control and it will create a package that does nothing but depend on the build dependencies.
[12:19:15] <seb_kuzminsky> and release 2.7.0~pre3, and i think it's time for 2.6.6 too...
[12:19:31] <seb_kuzminsky> mozmck: that's neat!
[12:19:49] <seb_kuzminsky> i usually run dpkg-checkbuilddeps and install the missing packages by hand
[12:20:11] <mozmck> very handy for setting up a new computer with the build dependencies for linuxcnc
[12:20:56] <mozmck> I'll work on this configure some. I am thinking of building a liveCD based on xubuntu 14.04
[12:21:06] <seb_kuzminsky> cool!
[12:21:11] <seb_kuzminsky> what kernel will you use?
[12:21:13] <mozmck> would you be interested in a buildbot for 14.04 packages?
[12:21:22] <seb_kuzminsky> very much so!
[12:21:36] <mozmck> We will be using preemt-rt, but I hope to build an rtai kernel as well.
[12:22:01] <mozmck> preemt-rt for us because that's all that will run the hm2_eth boards.
[12:22:08] <seb_kuzminsky> ubuntu does not ship a preempt-rt kernel for trusty, iirc
[12:22:14] <seb_kuzminsky> you'll be building your own?
[12:22:26] <mozmck> No, I'll probably build one from the stock kernel
[12:22:34] <mozmck> yes that is :)
[12:23:14] <seb_kuzminsky> this may be useful to you:
https://github.com/SebKuzminsky/linux-rtai-build
[12:23:37] <seb_kuzminsky> it's tbe build infrastructure i created for the rtai kernel we ship on wheezy (and precise, though no one cares about that i think)
[12:23:57] <mozmck> ok, thanks! I'll take a look at it.
[12:24:33] <seb_kuzminsky> it's got some good things: it uses the debian packaging of the kernel, and it does all the package-building inside a pbuilder chroot
[12:24:43] <mozmck> ah, nice.
[12:24:53] <mozmck> I need to learn pbuilder.
[12:30:22] <pcw_home> hm2_eth will run on Xenomai (but only Brand-M ATM)
[12:40:20] <skunkworks> and you need a supported nic
[12:40:27] <skunkworks> and run rt-net
[12:42:21] <pcw_home> Yep
[12:42:56] <pcw_home> Theres some hope that rt-net will be re-animated by the Xenomai group
[12:46:51] <skunkworks> cool
[13:03:53] <mozmck> So I guess the problem with rtai is that you would need rtai drivers for the NIC cards?
[13:06:10] <micges> mozmck: problem is that rtnet works under xenomai but hangs under rtai
[13:09:05] <pcw_home> RT-Net drivers worked at one time under RTAI but have apparently bit-rotted
[13:09:10] <mozmck> huh, so how does rtnet work?
[13:11:01] <micges> whole stack along with drivers are as realtime modules
[13:11:09] <pcw_home> Its a set of drivers and a real time network stack that supports basic comm (ARP/UDP/ICMP etc) and TDM
[13:11:25] <mozmck> so it has it's own drivers for NIC cards?
[13:12:11] <micges> yes
[13:12:30] <pcw_home> (I think the Xenomai people are looking at a the possibility of a wrapper around a normal driver)
[13:12:38] <skunkworks> <10 supported nics
[13:12:58] <mozmck> seb_kuzminsky: my tcl version is 8.6.1-4ubuntu1
[13:13:37] <skunkworks> sorry - 12 supported nics
[13:23:43] <KGB-linuxcnc> 03Francis Tisserant 05seb/master/po4a 425960f 06linuxcnc 10(27 files in 5 dirs) first french translation with po4a tool * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=425960f
[13:23:43] <KGB-linuxcnc> 03Francis Tisserant 05seb/master/po4a 7da37bd 06linuxcnc 03docs/src/getting-started/stepconf.fr.po 10docs/src/getting-started/stepconf.txt 10docs/src/getting-started/stepconf_fr.txt French doc po4a-translation * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=7da37bd
[13:23:43] <KGB-linuxcnc> 03Francis Tisserant 05seb/master/po4a 2e5a8b4 06linuxcnc 10(11 files) French doc, last po4a files from getting_started * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2e5a8b4
[13:24:24] <skunkworks> ooh!
[13:28:03] <skunkworks> It is all coming together!
[13:32:37] <seb_kuzminsky> awesome!
[13:33:01] <seb_kuzminsky> tiss was having trouble with the new signed-off-by policy, but i looks like he figured it out
[13:33:40] <seb_kuzminsky> this is still an experiment, but so far i see no show-stoppers and plenty of upside, and i think that's francis' opinion too
[13:39:13] <seb_kuzminsky> hmm, looks like he committed the _fr.txt files, that should not be done in that branch (they're generated from the english .txt and the french .po now)
[13:39:33] <seb_kuzminsky> i need to document the new workflow better
[15:44:48] <mozmck> seb_kuzminsky: I tried to rebase your new-deb-configure branch on master, but I think I did something wrong. I copied out the 3 modified files, did the rebase, and then tried to copy them back over the files git had mangled. This seemed to work but it looks like I lost the last couple of your commits.
[15:48:24] <seb_kuzminsky> hmm well that doesnt sound right
[15:48:51] <seb_kuzminsky> that's a weird rebase since it contains a whole-file replacement
[15:48:55] <seb_kuzminsky> so the 3diff is not very useful
[15:49:13] <seb_kuzminsky> let me try the rebase here, i'll push the updated branch, and you can look at that instead
[15:49:17] <mozmck> yes, which is why I saved the file out and just copied it back over.
[15:49:33] <mozmck> ok, would it make sense to do it on 2.7?
[15:50:13] <seb_kuzminsky> making a copy like that won't do the right thing either, since there have been changes (i bet) to debian/configure in master
[15:50:23] <seb_kuzminsky> and yeah, 2.7 is probably the right place.
[15:50:27] <seb_kuzminsky> though that's kind of scary
[15:50:44] <seb_kuzminsky> i wish i'd had the fortitude to finish that branch before i made the 2.7 branch like 6 months ago
[15:51:30] <seb_kuzminsky> i'll think about 2.7 vs master for that branch...
[15:51:36] <mozmck> there have been changes, but they don't matter really since it is bash to python etc. I was planning on having the files side by side to figure out stuff.
[15:52:46] <seb_kuzminsky> i would figure out the master commit that the new-deb-configure branch is based on, then git log debian/configure from that commit to origin/master, and manually transfer the intent of the changes to the new configure script
[15:53:36] <mozmck> oh, so that will show only commits that affect debian/configure?
[15:56:07] <seb_kuzminsky> yeah
[15:56:27] <seb_kuzminsky> something like: git log -p origin/seb/master/new-deb-configure..origin/master -- debian/configure
[15:56:30] <seb_kuzminsky> bbl
[17:43:04] <skunkworks> what are the odds of me find this in a random internet search... Maybe it isn't as big as we think..
[17:43:06] <skunkworks> http://ask.metafilter.com/95115/This-shouldnt-be-happening
[18:58:35] <seb_kuzminsky> skunkworks: a wild jepler appears!
[18:59:18] <seb_kuzminsky> skunkworks: i bumped into him elsewhere too just recently:
https://lists.debian.org/debian-devel/2015/02/msg00103.html
[18:59:40] <seb_kuzminsky> the dude is spreading knowledge all over the internet :-)
[19:23:12] <andypugh> It is annoying now, whenever I search for ner-a-car images half of them are mine.
[19:39:30] <Tom_itx> heh