#linuxcnc-devel | Logs for 2016-07-12

[00:03:46] <KimK> I like Debian on our CNC ISOs, and support the choice of Debian there. Keep things simple, great for a CNC machine or small PC where you don't need to do a lot of work. But Debian can be hard to live with for a long stay on the main desktop. I don't like Mint's "complicated parentage", and can see why you guys would prefer not to support it as developers (too many cooks?), but it sure is easy to live with on a long-term
[00:03:47] <KimK> basis. I like the way it works.
[00:03:50] <KimK> Bah!
[00:05:43] <KimK> Maybe I do need a character-counter, lol.
[00:07:04] <KimK> No, it's a new screen, I'll get used to it.
[07:01:51] <skunkworks> zlog
[07:32:42] <jepler> seb_kuzminsky: this time, jessie-rtpreempt-amd64 fell off the face of the planet, but differently than what wheezy did
[07:33:01] <jepler> [Failure instance: Traceback (failure with no frames): <class 'twisted.internet.error.ConnectionLost'>: Connection to the other side was lost in a non-clean fashion.
[07:33:04] <jepler> ]
[07:35:15] <jepler> my real wheezy amd64 vanilla system ran tests/and-or-not-mux.0 successfully 125k+ times overnight and of course my real jessie preempt amd64 system ran 25k+ yesterday so once again it smells like something that happens in vms, not real machines :-/
[08:37:07] <tinkerer> jepler: which kind of vms are you using?
[08:39:36] <jepler> tinkerer: seb_kuzminsky said last night, qemu/kvm
[08:41:03] <tinkerer> ok, then it's unlikely that it is this issue: http://serverfault.com/a/608441
[08:43:00] <jepler> most of the problems in the past 2 days have been "rcu_preempt detected stall"
[08:43:08] <jepler> this last one is different
[08:43:16] <jepler> afk
[08:45:48] <tinkerer> ipv6 issues?
[09:18:30] <seb_kuzminsky> jepler: black console, totally unresponsive, bounced it
[09:19:06] <seb_kuzminsky> tinkerer: no ipv6 is used in the buildbot
[09:32:10] <tinkerer> and configured?
[09:36:56] <tinkerer> sudo modprobe -c | grep ipv6
[09:40:50] <linuxcnc-build> build #2092 of 3003.dsc-wheezy is complete: Failure [4failed making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3003.dsc-wheezy/builds/2092
[09:41:38] <linuxcnc-build> build #2218 of 3301.dsc-wheezy-rtpreempt is complete: Failure [4failed making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3301.dsc-wheezy-rtpreempt/builds/2218
[09:41:53] <linuxcnc-build> build #763 of 3004.dsc-jessie is complete: Failure [4failed making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3004.dsc-jessie/builds/763
[09:42:17] <linuxcnc-build> build #762 of 3302.dsc-jessie-rtpreempt is complete: Failure [4failed making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3302.dsc-jessie-rtpreempt/builds/762
[09:42:56] <linuxcnc-build> build #3514 of 3002.dsc-precise is complete: Failure [4failed making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3002.dsc-precise/builds/3514
[09:48:26] <linuxcnc-build> build #4351 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4351
[09:52:02] <jepler> sigh one of these days I'll get the debian stuff right
[11:06:26] <skunkworks> http://www.machsupport.com/forum/index.php/topic,32854.0.html
[11:06:32] <skunkworks> http://www.machsupport.com/forum/index.php/topic,32854.0.html
[11:06:58] <skunkworks> sorry - meant
[11:06:59] <skunkworks> http://www.machsupport.com/forum/index.php/topic,32833.0.html
[11:07:05] * skunkworks hugs linuxcnc
[11:07:26] <seb_kuzminsky> group hug!
[11:07:35] <skunkworks> Mmmmmm
[11:08:24] <skunkworks> sorry - I lingered
[11:13:22] <cradek> I suppose like threading, every external device has to implement probing
[11:14:34] <seb_kuzminsky> has anyone made an arm box with linuxcnc to sell to the mach people yet?
[11:14:55] <cradek> I thought alex made a cute self-contained thing
[11:35:44] <jepler> yes I think he did
[11:36:15] <jepler> seb_kuzminsky: this again! [Failure instance: Traceback (failure with no frames): <class 'twisted.internet.error.ConnectionLost'>: Connection to the other side was lost in a non-clean fashion.
[11:36:23] <jepler> http://buildbot.linuxcnc.org/buildbot/builders/1530.rip-jessie-rtpreempt-amd64/builds/967/steps/runtests
[11:37:40] <jepler> coincidence? it was during the flipflop.0 test both times
[11:41:28] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus a2ab3ae 06linuxcnc 10(8 files) packaging: rtai, xenomai are sub-packages of uspace * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a2ab3ae
[11:41:28] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus bdc41c5 06linuxcnc 10docs/src/code/building-linuxcnc.txt docs: document new RTOS support * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=bdc41c5
[11:43:43] <jepler> struct timespec ts = {0, 10000000};
[11:43:44] <jepler> rtapi_clock_nanosleep(CLOCK_MONOTONIC, 0, &ts, NULL, NULL);
[11:43:57] <jepler> that last failure is unfortunately after I changed the message-printing thread to have a much larger sleep
[11:44:05] <jepler> so it blows that theory out of the water
[11:49:15] <skunkworks> http://electronicsam.com/images/matsuura/DSC_7726.jpg
[11:53:31] <jepler> what did you do, double-sided-tape the LCD to the old control?
[11:57:32] <seb_kuzminsky> skunkworks: does it have a standard vesa mount on the back?
[11:57:56] <skunkworks> yes
[11:58:06] <skunkworks> 4 holes
[11:58:11] <seb_kuzminsky> nice
[11:58:29] <seb_kuzminsky> i'm been (cough) meaning to do something similar on my bridgeport, for oh about the past 5 years now
[11:59:01] <jepler> seb_kuzminsky: ^^^ I keep breaking buildbot
[11:59:07] <seb_kuzminsky> jepler: same symptoms, black console, unresponsive to the keyboard
[11:59:36] <seb_kuzminsky> pid 26238 is the stalled one, in case that helps any
[11:59:54] <seb_kuzminsky> it unblanked when i hit the power button, but it doesnt seem to be shutting down
[11:59:56] <jepler> umm not sure how a pid would help
[12:00:08] <seb_kuzminsky> if it got logged somewhere, somehow
[12:00:14] <seb_kuzminsky> imma power cycle it
[12:02:16] <jepler> maybe I should reboot my jessie machine with maxcpus=2 or =1 and see if that lets me produce a problem
[12:03:28] <jepler> of course it could also be kernel-dependent
[12:04:00] <jepler> seb_kuzminsky: which kernel is on the jessie builder?
[12:04:37] <seb_kuzminsky> it's 4.1.0-0.bpo.1-rt-amd64
[12:04:39] <seb_kuzminsky> http://buildbot.linuxcnc.org/buildbot/builders/1530.rip-jessie-rtpreempt-amd64/builds/967/steps/environment-report/logs/stdio
[12:04:59] <seb_kuzminsky> i should probably upgrade it
[12:05:15] <seb_kuzminsky> would that mess up your testing?
[12:06:24] <jepler> if you upgrade it and the problem goes away that's dandy
[12:06:28] <jepler> you'd put that current 4.6 kernel in?
[12:06:39] <jepler> .. but wasn't 4.1 the one we were talking about standardizing on?
[12:06:47] <seb_kuzminsky> how about if i upgrade it and it changes everything so you get different errors? :-P
[12:07:00] <jepler> I've been testing with 4.4 but I'll reboot with 4.1 and see how it goes
[12:12:05] <jepler> 700+ iterations of flipflop.0 with Linux rat 4.1.0-0.bpo.2-rt-amd64 #1 SMP PREEMPT RT Debian 4.1.6-1~bpo8+1 (2015-09-09) x86_64 GNU/Linux
[12:12:16] <jepler> on bare metal with 8 CPUs
[12:12:43] <seb_kuzminsky> i should try that kernel on the buildbot
[12:14:18] <jepler> I wish it wasn't such a PITA to get packages from snapshot.debian.org!@
[12:15:48] <jepler> hmm we should move hm2_test into its test, and make sure a new hostmot2 hardware driver can be built just with exported headers (they can't be right now)
[12:16:23] <jepler> huh this is bizarre
[12:17:12] <jepler> "1" is missing from the font that xfce uses for most of its user interface
[12:17:33] <seb_kuzminsky> you can get by with just "0"
[12:17:35] <seb_kuzminsky> and inc
[12:17:39] <jepler> so e.g., the clock reads " :50: 9 20 6-07- 2" (11:50:19 2016-07-12)
[12:19:04] <jepler> it's fixed if I change the defult font from "sans 0" (er "sans 10") to "sans 9"!
[12:19:47] <jepler> or turn off subpixel fonts or change hinting to "slight" or "none"
[12:20:18] <mozmck> That sounds really odd. Seems like something that major would have been noticed and fixed?
[12:20:38] <jepler> It must not happen all the time, I would have noticed it before now
[12:22:11] <seb_kuzminsky> Linux jessie-rtpreempt-i386 4.1.0-2-rt-686-pae #1 SMP PREEMPT RT Debian 4.1.6-1 (2015-08-23) i686 GNU/Linux
[12:22:15] <linuxcnc-build> build #2093 of 3003.dsc-wheezy is complete: Failure [4failed making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3003.dsc-wheezy/builds/2093 blamelist: dummy, Jeff Epler <jepler@unpythonic.net>
[12:22:28] <seb_kuzminsky> both jessie preempt-rt machines will soon be running the latest 4.1-rt kernel from snapshots
[12:22:44] <seb_kuzminsky> linux-image-4.1.0-2-rt-amd64_4.1.6-1_amd64.deb
[12:22:51] <linuxcnc-build> build #2219 of 3301.dsc-wheezy-rtpreempt is complete: Failure [4failed making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3301.dsc-wheezy-rtpreempt/builds/2219 blamelist: dummy, Jeff Epler <jepler@unpythonic.net>
[12:22:59] <jepler> I rebooted into this kernel with kexec, I bet I am hitting this and it makes the graphics stack behave a bit weirdly: https://bugs.freedesktop.org/show_bug.cgi?id=95023
[12:23:11] <linuxcnc-build> build #764 of 3004.dsc-jessie is complete: Failure [4failed making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3004.dsc-jessie/builds/764 blamelist: dummy, Jeff Epler <jepler@unpythonic.net>
[12:23:23] <linuxcnc-build> build #763 of 3302.dsc-jessie-rtpreempt is complete: Failure [4failed apt-get-update install-missing-build-dependencies making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3302.dsc-jessie-rtpreempt/builds/763 blamelist: dummy, Jeff Epler <jepler@unpythonic.net>
[12:24:00] <seb_kuzminsky> jepler: nice find
[12:24:17] <seb_kuzminsky> what's the branch with the freebsd build fixes?
[12:24:18] <linuxcnc-build> build #3515 of 3002.dsc-precise is complete: Failure [4failed making debian source package] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/3002.dsc-precise/builds/3515 blamelist: dummy, Jeff Epler <jepler@unpythonic.net>
[12:24:40] <jepler> seb_kuzminsky: some stuff is in master, for the rest look at github PRs
[12:24:55] <seb_kuzminsky> ok
[12:25:11] <jepler> I know trasz has some more that look cherry-pick-able, and then I have a couple of "see if this works for you" type PRs that I would like his feedback on
[12:29:06] <jepler> linuxcnc-build: dance
[12:29:07] <linuxcnc-build> <(^.^<)
[12:29:08] <linuxcnc-build> <(^.^)>
[12:29:09] <linuxcnc-build> (>^.^)>
[12:29:10] <linuxcnc-build> (7^.^)7
[12:29:11] <linuxcnc-build> (>^.^<)
[12:30:45] <jepler> ugh so bored waiting for these long running mdi tests
[12:31:02] <linuxcnc-build> build #4352 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4352 blamelist: dummy, Jeff Epler <jepler@unpythonic.net>
[12:31:05] <seb_kuzminsky> jepler: ok, both jessie-rtpreempt buildslaves are running the latest 4.1-rt kernel now
[12:37:47] <jepler> the vm hangs / crashes make me worry there's something subtly broken about uspace
[12:37:55] <jepler> I wish someone smart would look at it
[12:38:18] <mozmck> So is 4.1 best of the newer kernels? I know PCW had some problems with 4.something on one board anyhow.
[12:39:07] <jepler> mozmck: 4.1 seemed to have the best compatibility, but PCW has a least one board where 4.6 (I think) is better. Overall our guess is that 4.1 had better compatibility, though.
[13:10:38] <jepler> yay(?) reproduced the hang on bare metal with maxcpus=2
[13:10:44] <seb_kuzminsky> yay
[13:10:50] <cradek> oh good
[13:10:50] <seb_kuzminsky> definily
[13:10:57] <cradek> buildbot can live to see another minute or two
[13:12:27] <jepler> I assume bisect is going to point me at the commit that changed how messages are exfiltrated from realtime, but maybe the answer will surprise me..
[13:12:53] <jepler> at origin/mster survived 800+ iterations, while uspace-plus was blowing up well under 100.
[13:14:03] <jepler> [ 225.330287] rtapi_app[12940]: segfault at 0 ip 000000000040d78c sp 00007febc6ae5eb8 error 4 in rtapi_app[400000+18000]
[13:14:18] <seb_kuzminsky> that's different
[13:14:27] <jepler> yeah it is
[13:21:09] <jepler> OK I understand this crash, it's a bug I've introduced and then (hopefully!) fixed in my tree
[13:22:11] <jepler> at 669dca9d I arranged to destroy the App object and set the app pointer to NULL
[13:22:31] <jepler> but it wasn't until 25a055cd that I ensured all tasks were stopped before doing that
[13:22:48] <jepler> the second commit that bisect gave me, 36fe2786, is in between those two
[13:23:09] <jepler> 669dca9 uspace: arrange to destroy the RtapiApp at shutdown
[13:23:17] <jepler> 36fe278 uspace: Introduce lockfree queue for rtapi_print_msg
[13:23:20] <jepler> 25a055c uspace: ensure a more orderly shutdown
[13:24:14] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.7 5e19f08 06linuxcnc 10src/configure.in configure: fix a typo in a hep message * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=5e19f08
[13:40:18] <jepler> hm, no, bisect is bad before the commit I suspected..
[13:44:47] <jepler> uspace: stop threads and then destroy the RtapiApp at shutdown
[13:44:53] <jepler> .. is the first bad commit
[13:45:03] <jepler> that's the combination of 669dca9 and 25a055c from above
[14:15:07] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15jepler opened issue #109: figure out why trying to clean up tasks crashes uspace posix with maxcpus=2 02https://github.com/LinuxCNC/linuxcnc/issues/109
[14:46:22] <jepler> hmph
[14:46:38] <jepler> with maxcpus=2, xenomai fails about 1 time in 50 running this script with halrun:
[14:46:41] <jepler> loadrt threads name1=fast period1=1000000
[14:46:43] <jepler> dmesg says
[14:46:43] <jepler> Jul 12 14:19:04 rat kernel: [ 1507.004904] Xenomai: watchdog triggered -- signaling runaway thread 'rtapi_app'
[14:46:47] <jepler> start
[14:46:49] <jepler> loadusr -w sleep .1
[15:11:00] <kwallace> Yay, the laptop stork left a package on the doorstep. It came with the hard drive door (but no caddy) and a power supply. It booted right up, with Windows 7, whatever that is.
[15:21:08] <skunkworks> zlog
[15:22:18] <jepler> kwallace: as long as you can whip up something that keeps the drive snug and doesn't add too much thermal insulation, you'll be fine. I've resorted to carefully positioned rubber bands in my life.
[15:23:59] <andypugh> “You guys suck because my pathological INI file breaks the conversion script”
[15:26:12] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 88be437 06linuxcnc 10src/rtapi/rtapi_uspace.hh 10src/rtapi/uspace_common.h 10src/rtapi/uspace_rtapi_app.cc 10src/rtapi/uspace_ulapi.c uspace: rtapi_get_time will need a different implementation for rtai * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=88be437
[15:26:13] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 8dbccfb 06linuxcnc 10(40 files in 7 dirs) build: include a copy of boost lockfree for heritage platforms * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8dbccfb
[15:26:13] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus f0ae455 06linuxcnc 10src/rtapi/uspace_common.h 10src/rtapi/uspace_rtapi_app.cc 10src/rtapi/uspace_ulapi.c uspace: Introduce lockfree queue for rtapi_print_msg * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=f0ae455
[15:26:16] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 3c0e6ea 06linuxcnc 10src/rtapi/rtapi_uspace.hh 10src/rtapi/uspace_common.h 10src/rtapi/uspace_rtapi_app.cc 10src/rtapi/uspace_ulapi.c uspace: rtapi_delay will need a different implementation for rtai * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=3c0e6ea
[15:26:21] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 4841a94 06linuxcnc 10src/configure.in configure: fall back to uspace realtime if rtai not found * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=4841a94
[15:26:25] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 6e4ad10 06linuxcnc 10src/module_helper/module_helper.c module_helper: Allow loading modules associated with running kernel * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=6e4ad10
[15:26:29] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 7a809ac 06linuxcnc 10src/module_helper/module_helper.c module_helper: always allow .ko objects * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=7a809ac
[15:26:34] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 14e582d 06linuxcnc 10scripts/realtime.in realtime: Load modules even for uspace, if requested by rtapi.conf * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=14e582d
[15:26:37] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 443bbd1 06linuxcnc 10src/Makefile build: even on uspace, make linuxcnc_module_helper setuid * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=443bbd1
[15:26:39] <skunkworks> andypugh, you can't please anyone.
[15:26:41] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus b67d1f4 06linuxcnc 10src/Makefile build: install additional programs as setuid in uspace * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=b67d1f4
[15:26:45] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus ab4ade8 06linuxcnc 10debian/configure packaging: drop special treatment of linuxcnc_module_helper * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ab4ade8
[15:26:49] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 81b6744 06linuxcnc 10(7 files in 3 dirs) uspace: add uspace+rtai realtime * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=81b6744
[15:26:53] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 4b46b5d 06linuxcnc 10(6 files in 2 dirs) uspace: add uspace+xenomai realtime * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=4b46b5d
[15:26:57] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 013facf 06linuxcnc 10(8 files) packaging: rtai, xenomai are sub-packages of uspace * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=013facf
[15:27:01] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus 434c782 06linuxcnc 10docs/src/code/building-linuxcnc.txt docs: document new RTOS support * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=434c782
[15:27:07] <andypugh> Why did he comment-out the [HEADER] part?
[15:27:21] <kwallace> jepler, I think I know somebody that can make something on a CNC that will fit.
[15:27:44] <skunkworks> andypugh, the carousel component is working great - I have not had a chance to look at the jogging issue.
[16:29:11] <linuxcnc-build> build #484 of 4017.5.deb-wheezy-armhf is complete: Failure [4failed shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4017.5.deb-wheezy-armhf/builds/484 blamelist: dummy, Jeff Epler <jepler@unpythonic.net>
[16:31:12] <jepler> seb_kuzminsky: have you noticed that the deb builders for build N can be going at the same time as the RIP builders for build N+1? feature or bug?
[16:59:36] <seb_kuzminsky> jepler: intentional
[16:59:41] <seb_kuzminsky> though maybe misguided
[17:00:05] <seb_kuzminsky> the thinking was, as long as only one thread runs runtests, it's just system load and that should be ok
[17:00:32] <seb_kuzminsky> but you know, i've been wrong before
[17:00:34] <seb_kuzminsky> once
[17:00:46] <seb_kuzminsky> and that was when i thought i was wrong about something, but actually i was right
[17:07:48] <trasz> seb_kuzminsky: thanks for the fixes :-)
[17:07:57] <seb_kuzminsky> sure :-)
[17:08:07] <trasz> seb_kuzminsky: they are better than mine, but if you're doing some more work, you might want to wait a day or two.
[17:08:24] <trasz> seb_kuzminsky: i basically have all this building and working; it's just that i need to clean things up and push upstream.
[17:08:25] <seb_kuzminsky> i'll wait
[17:08:34] <seb_kuzminsky> oh that's great to hear
[17:09:09] <trasz> seb_kuzminsky: plan for now is to get the port committed to the ports collection - it was waiting for tkimg update, which finally happened today.
[17:09:27] <trasz> seb_kuzminsky: then i'll go back to submitting fixes to linuxcnc.
[17:09:46] <seb_kuzminsky> is something wrong with the libimg port?
[17:09:56] <seb_kuzminsky> or, was
[17:09:58] <trasz> seb_kuzminsky: there was; it didn't work correctly with libpng.
[17:10:18] <trasz> seb_kuzminsky: which made axis crash at startup.
[17:10:23] <seb_kuzminsky> heh yikes
[17:10:46] <seb_kuzminsky> my freebsd 10 vm is about set up, when you push your stuff i'll be able to test it
[17:11:10] <seb_kuzminsky> have you run the latency tests? i'm curious how a stock freebsd kernel does
[17:11:47] <trasz> seb_kuzminsky: it does horribly, but it still doesn't use the proper scheduling.
[17:12:05] <trasz> seb_kuzminsky: i've tried enabling it and it hang; didn't have time to track this down yet.
[17:12:22] <trasz> seb_kuzminsky: i'm wondering myself; in theory freebsd should do well.
[17:12:32] <trasz> seb_kuzminsky: due to interrupt threads, priority lending etc.
[17:12:32] <seb_kuzminsky> does freebsd have anything better than posix 1003.1?
[17:12:55] <trasz> seb_kuzminsky: no, but the plain SCHED_RR didn't work for some reason.
[17:13:04] <trasz> seb_kuzminsky: but it should; there must be a bug somewhere.
[17:13:19] <seb_kuzminsky> right
[17:14:34] <trasz> seb_kuzminsky: ah, that reminds me.
[17:14:48] <trasz> seb_kuzminsky: can you tell me which parts of linuxcnc use readline?
[17:15:06] <trasz> seb_kuzminsky: another thing i need to sort out is to try to link it against libedit, which provides readline-compatible API.
[17:15:18] <trasz> seb_kuzminsky: this would fix the licensing conflict, allowing to distribute binary packages.
[17:15:42] <seb_kuzminsky> i dont know the list of readline users off the top of my head, i'd have to go digging in the build system
[17:16:14] <trasz> seb_kuzminsky: ok, never mind then; i'll need to go through this anyway.
[17:17:13] <seb_kuzminsky> looks like halcmd, sai, and gladevcp, at first glance
[17:19:56] <trasz> thanks
[17:20:07] <trasz> ok, but that's for tomorrow.
[17:20:17] <trasz> :-)
[17:20:47] <seb_kuzminsky> sounds good
[17:26:44] <jepler> trasz: hi, sorry i've been neglecting your most recent pr. other stuff has been foremost on my mind...
[17:28:03] <jepler> I hope my uspace work for multiple linux rtoses doesn't make too many conflicts
[17:31:07] <jepler> gladevcp surprises me as a possible readline user.
[17:36:09] <seb_kuzminsky> oh, i think i'm confused about gladevcp, i think it just calls python's file-object readlines() method
[17:36:21] * seb_kuzminsky crawls back into his sewer pipe
[17:54:59] <andypugh> Testing this comversion script is really tedious. I have to keep re-setting the config to non-JA. I suppose there is probably a cunnign way to set a restore point with git to make it less tedious.
[18:55:55] <seb_kuzminsky> andypugh: using git for that is absolutely a good way to do it
[18:56:29] <JT-Shop> can you say git reset --hard
[18:56:36] <seb_kuzminsky> 'git commit' all the configs you want to test, run your converter on all of them and check the results, then run 'git reset --hard' to restore the files to their pre-converted state
[18:56:50] <seb_kuzminsky> and 'git clean -fdx .' to get rid of all the generated, untracked files
[18:56:52] <seb_kuzminsky> JT-Shop: +1
[18:57:10] <JT-Shop> a new one for me git glean -fdx
[18:57:15] <JT-Shop> what is -fdx
[18:57:27] <seb_kuzminsky> force, remove dirs, and i forget
[18:57:42] <JT-Shop> which one is i forget?
[18:57:50] <JT-Shop> just kidding
[18:57:50] <seb_kuzminsky> oh yeah, even remove files that are ignored by .gitignore
[18:58:54] <JT-Shop> and added to my git-secrets.txt file
[18:59:53] <seb_kuzminsky> i hope it says The Secret Life of Gits
[19:00:26] <JT-Shop> yep
[19:02:03] <seb_kuzminsky> back when i was working at the university i taught a short class called Habits of Highly Effective Unix
[19:03:15] <JT-Shop> I used to like the show secret life of machines
[20:21:28] <linuxcnc-build> build #486 of 4017.5.deb-wheezy-armhf is complete: Failure [4failed shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4017.5.deb-wheezy-armhf/builds/486 blamelist: dummy, Jeff Epler <jepler@unpythonic.net>
[20:22:11] <skunkworks> http://electronicsam.com/images/matsuura/DSC_7732.jpg
[20:28:41] <seb_kuzminsky> shiny!
[20:32:15] <seb_kuzminsky> my conversion is so old it still says EMC2 on it: http://highlab.com/~seb/linuxcnc/bp/2010.08-control-panel/IMG_20160627_211813.jpg
[20:32:34] <skunkworks> ncie!
[20:32:36] <skunkworks> nice
[20:32:47] <seb_kuzminsky> PCIe!
[20:32:49] <seb_kuzminsky> NVME!
[20:34:54] <skunkworks> There are still a few things to hook up. over current loops, lube level, some random stuff that needs to be checked. but the doors close now and it runs well. Still need to make it gracefully recover from a failed tool change. No rush
[20:35:07] <skunkworks> (hitting the estop while it is changing...)
[20:35:36] <seb_kuzminsky> your machines always look so nice and finished - very impressive
[20:39:36] <KGB-linuxcnc> 03Sebastian Kuzminsky 05master 3e39257 06linuxcnc 10configs/sim/axis/gantry.ini sim/axis/gantry config: fix extra A axis from ini update to JA * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=3e39257
[20:39:37] <KGB-linuxcnc> 03Sebastian Kuzminsky 05master 635339e 06linuxcnc 10src/emc/motion/command.c Motion: Abort doesn't take a joint number argument * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=635339e
[20:59:38] <cradek> mine's so old it may still be running emc2: http://timeguy.com/cradek-files/emc/jr.jpg
[21:01:37] <skunkworks> cradek, yours is much nicer :)
[21:02:59] <cradek> I didn't even label the switches
[21:03:07] <cradek> I just poke them and see what happens if I forget
[21:03:34] <skunkworks> heh - so far ours are just sharpied on
[21:04:01] <cradek> bet sharpie isn't coolant-proof
[21:04:51] <cradek> so if a bill is due today and I put it in the mailbox tonight, that's good enough, right?
[21:08:01] <seb_kuzminsky> i love how you had to built a hat into the ceiling to make JR fit
[21:08:12] * seb_kuzminsky is working on 2.7.5
[21:08:17] <cradek> whee!
[21:08:50] <cradek> it *almost* fit without it...
[21:09:08] <seb_kuzminsky> > git log --oneline v2.7.4..origin/2.7 | wc -l
[21:09:09] <seb_kuzminsky> 141
[21:09:19] <cradek> jeez
[21:11:04] <seb_kuzminsky> my 2.7.5-prep branch has commits from 2 months ago...
[21:18:16] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus f03c756 06linuxcnc 10(8 files) packaging: rtai, xenomai are sub-packages of uspace * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=f03c756
[21:18:16] <KGB-linuxcnc> 03Jeff Epler 05jepler/master/uspace-plus b9b6edf 06linuxcnc 10docs/src/code/building-linuxcnc.txt docs: document new RTOS support * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=b9b6edf
[22:35:56] <jepler> a bunch of new -rc1 rt kernel patches out today, from 3.2 to 4.4
[22:41:29] <seb_kuzminsky> just when i thought i was getting caught up
[22:42:43] <jepler> and there's a new x.y.0 release of joomla for the g-d forum
[22:48:41] <jepler> well hopefully that didn't run anything. goodnight.
[23:32:51] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.7 383fb03 06linuxcnc 10VERSION 10debian/changelog LinuxCNC 2.7.5 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=383fb03
[23:32:51] <KGB-linuxcnc> 03Sebastian Kuzminsky 05signed tags 7f680c5 06linuxcnc 03v2.7.5 LinuxCNC v2.7.5 (tagged commit: 383fb03) * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=7f680c5