#linuxcnc-devel | Logs for 2014-10-09

[00:53:54] <seb_kuzminsky> :q
[09:30:32] <KGB-linuxcnc> 03Dewey Garrett 05master 4ef6ce7 06linuxcnc 10(5 files in 3 dirs) inihal: add hal pins for trajectory planner items * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=4ef6ce7
[09:30:32] <KGB-linuxcnc> 03Dewey Garrett 05master 2ab3a91 06linuxcnc 10src/emc/ini/inihal.cc inihal.cc: use %g for reals in Changed: prints * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2ab3a91
[09:43:17] <seb_kuzminsky> brianmorel99: thanks for looking into the problem, i'll be happy to help you wade through the kernel build system we use
[09:43:25] <seb_kuzminsky> (and even happier to hand off kernel-building to you!)
[09:46:20] <cradek> that usb thing is the only problem we know about with our kernel - it would be great to have it fixed
[09:50:07] <seb_kuzminsky> dgarr: i used inihal the other day to tune accel & vel settings for a machine i'm building for a friend, it was great
[09:50:20] <seb_kuzminsky> way more convenient than restarting linuxcnc for each tweak
[09:50:27] <seb_kuzminsky> cradek: that's the only bug i know of too
[09:51:08] <seb_kuzminsky> oh wait, no
[09:51:25] <cradek> argh
[09:51:31] <seb_kuzminsky> dgarr has a machine that worked fine with the lucid 2.6.32 rtai kernel but doesnt work with the 3.4.55 rtai kernel
[09:51:45] <cradek> oh right
[09:51:50] <cradek> it was a rtai regression
[09:52:07] <seb_kuzminsky> yeah
[10:50:37] <brianmorel99> I think the linked commit should fix it, but I won't be able to test until I get home tonight. http://goo.gl/C45gzB
[10:51:31] <brianmorel99> Reading the datasheet for the J1800, etc. it looks like the USB ports are pinmuxed to both the ehci & xhci controller and there is a problem when they both try to claim the ports.
[10:51:55] <brianmorel99> I'm still going to try the stock debian live-cd to see if I just missed the errors when I installed that.
[10:52:34] <seb_kuzminsky> brianmorel99: awesome!
[10:53:06] <seb_kuzminsky> i think skunkworks tried the debian wheezy live-cd (from upstream, not from linuxcnc) and reported that it worked.
[10:53:28] <seb_kuzminsky> that live-cd uses debian's 3.2 kernel, maybe there was a commit between 3.2 and 3.4.55 that broke it
[11:01:42] <brianmorel99> I was going to browse the kernel source for the debian 3.2 kernel today at work, but their build system makes my head spin (svn vs. git & patches vs. a full tree).
[11:03:45] <seb_kuzminsky> haha yeah
[11:11:47] <pcw_home_> brianmorel99: its great you are looking into the Baytrail USB issue, the J1800/1900/2900 MBs are very nice Linuxcnc platforms
[14:50:27] <cradek> that whole "we should deprecate I/O pins" was actually a disguised question "how do I sets my index-enable to test my encoder?" ???
[14:50:34] <cradek> good grief
[14:51:52] <seb_kuzminsky> heh yeah
[14:52:08] <seb_kuzminsky> are there any IO pins other than index-enable?
[14:52:54] <cradek> possibly, but not that you commonly encounter
[14:53:47] <seb_kuzminsky> yeah there are a bunch
[14:54:48] <seb_kuzminsky> including many that probably shouldnt be IO
[14:54:58] * seb_kuzminsky sighs and closes that window
[14:58:48] <cradek> ugh
[14:59:08] <cradek> well we should deprecate using IO pins where it's inappropriate to use IO pins
[14:59:26] <seb_kuzminsky> haha
[15:00:13] <cradek> upon further consideration, I propose that we should deprecate doing all inappropriate things
[15:03:29] <jepler> cradek: did you read back to the discussion of merge commits and SOB?
[15:03:36] <jepler> I put seb_kuzminsky off by saying I wanted more opinions about the matter
[15:04:58] <cradek> I think I read it the first time through
[15:05:08] <jepler> ah but you kept mum
[15:05:18] <cradek> does it take an extra step that's unlike what any (or most) other projects have?
[15:06:19] <jepler> git and linux kernel don't seem to sign off merge commits
[15:06:39] <jepler> and as seb says, there's no "git merge -s" to add the sign-off message to a merge commit
[15:08:07] <cradek> what's the risk of not requiring it? that someone acting in bad faith will put significant changes (that he doesn't want to SOB) in the merge?
[15:10:09] <jepler> yes
[15:10:57] <cradek> eh, that's just a fool us once shame on you problem
[15:11:11] <cradek> I think we shouldn't make day-to-day difficult to avoid those
[15:11:17] <jepler> OK
[15:11:29] <jepler> well I think you just need to add --no-merges to git rev-list to skip them
[15:12:52] <seb_kuzminsky> thanks guys
[15:13:20] <cradek> done
[15:13:28] <jepler> (don't you hate it when people give you prose, not a patch?)
[15:13:40] <cradek> for one line, I can handle it
[15:14:15] <brianmorel99> seb: Is your kernel builder able to add a couple of patches fairly easily ?
[15:19:48] <seb_kuzminsky> it's pretty much the debian project's kernel builder, so yeah it handles patches well
[15:21:21] <seb_kuzminsky> you'd add the patch file to the proper place in debian/patches/bugfix/, then add it to debian/patches/series, then rebuild and test
[15:26:31] <brianmorel99> Ok, I found a couple patches that were added into their 3.2 that deal with the handling of INTx on the PCI bus. Supposedly they fix the Bay Trail USB issue. When I get home tonight I'll fire up a VM and see if I can duplicate your setup.
[15:31:53] <seb_kuzminsky> brianmorel99: aaaaaawesome
[15:32:37] <seb_kuzminsky> i'll be near to irc tonight in case you cut yourself on any of the many sharp corners of my build monstrocity
[15:34:40] <brianmorel99> Well, it may be a process, there is laundry and a Red Wings season opener tonight, so not much may get done.
[15:38:07] <seb_kuzminsky> i am familiar with this pattern of development
[15:39:25] <seb_kuzminsky> my plan for tonight is to get pid control of the heated print bed in my linuxcnc 3d printer, using a Teensy for temperature sensing and a RAMPS for driving the resistive heater
[15:40:16] <seb_kuzminsky> instead of working on bug #394, which would be the responsible thing to do... :-/
[15:41:29] <jepler> seb_kuzminsky: it does sound like fun
[15:44:01] <seb_kuzminsky> i had to extend your arduino/hal protocol a bit to get access to all the bits on the analog channels on the teensy
[15:45:16] <seb_kuzminsky> it'd be good to have a 3d printer example in the sample configs
[19:06:35] <brianmorel99_> seb: http://pastebin.com/yMTFHATC
[19:07:16] <brianmorel99_> Red Wings are starting, so I won't dig into it for a bit. just posting in case you recognize something. It's a 14.04 Ubuntu w/ Python 2.7.6
[19:19:50] <jepler> whatever program gencontrol.py is trying to execute is not installed on your system
[19:20:04] <jepler> the traceback ending at around line 32 doesn't show what program it was...
[19:22:44] <jepler> that gencontrol.py doesn't seem to match the one in seb's github, https://github.com/SebKuzminsky/linux-rtai-debian/blob/3.4.55-rtai/lib/python/debian_linux/gencontrol.py
[19:23:10] <jepler> because do_arch_packages starts at line 170, and has a trivial body ('pass')
[19:24:52] <jepler> it's possible the program in question is kernel-wedge from package kernel-wedge. try installing that program and running the process again.
[19:25:12] <jepler> brianmorel99_: ^^ just some idle speculation
[19:46:25] <brianmorel99_> First intermission
[19:46:50] <brianmorel99_> jepler: I think your right, my kernel-wedge build failed and I didn't catch it.
[20:46:37] <seb_kuzminsky> brianmorel99, brianmorel99_: i've never tried to build on trusty, i did my builds on wheezy, but that shouldnt matter
[20:46:43] <seb_kuzminsky> since the builds happen in pbuilder
[20:47:22] <brianmorel99_> i get an error because some packages being downloaded couldn't be verified. I'm searching for key id's now
[20:47:27] <seb_kuzminsky> did you see in the README the "Using this build system" section?
[20:47:47] <seb_kuzminsky> err oh yeah, i probably spaced putting the key fetching instructions in the readme
[20:47:53] <seb_kuzminsky> if you add them, send me a PR and i'll merge it
[20:48:02] <seb_kuzminsky> the key ids are in the top-level makefile
[20:48:25] <seb_kuzminsky> sudo apt-key adv --keyserver hkp://keys.gnupg.net --recv-key $ID
[20:48:38] <brianmorel99_> ok, thanks
[20:48:52] <seb_kuzminsky> there are probably a bunch of build dependencies that are not recorded anywhere
[20:51:04] <brianmorel99_> ya, I've been fixing those as I go.
[20:53:21] <seb_kuzminsky> if you send me a list i'll add it to the readme
[20:53:31] <seb_kuzminsky> (or send me a pr which fixes the readme)
[22:01:23] <brianmorel99_> I think I'm running into an issue with not signing the packages I create. Too tired now, try again tomorrow.
[22:38:39] <seb_kuzminsky> brianmorel99: in the top-level makefile, near the top, there's a variable called 'ARCHIVE_SIGNING_KEY', which should identify the key you want to use to sign the deb archive that the build system builds
[22:38:47] <seb_kuzminsky> but that's great, it means you're getting really close