#linuxcnc-devel | Logs for 2016-05-20

Back
[00:04:37] <KGB-linuxcnc> 03chris morley 05master 5c1bad8 06linuxcnc 10lib/python/gladevcp/tooledit_widget.py gladevcp: add functions to change title and tab font * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=5c1bad8
[00:04:37] <KGB-linuxcnc> 03chris morley 05master edcaf9f 06linuxcnc 10docs/src/gui/gladevcp.txt docs: update gladevcp tooledit widget doumentation * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=edcaf9f
[07:40:51] <jepler> .. my little usb-ethernet test went south overnight. from the kernel log, it appears the dongle reported the interface state changed from down to up multiple times overnight, so obviously that's not going to work out.
[07:42:21] <jepler> boy even on this desktop you feel the fan slow down as soon as you exit realtime
[07:47:31] <tinkerer> Hi
[07:48:51] <tinkerer> are there some interests on a working image for the raspi2 with lcnc and hm2 spi driver?
[07:52:03] <tinkerer> at the moment it works with machinekit
[08:32:47] <jepler> tinkerer: we would be happy to take working hardware drivers in our tree.
[08:33:17] <jepler> I think producing a whole installer image for a new piece of hardware is not something we can take on without a developer dedicated to doing it.
[08:48:32] <pcw_home> jepler: I tried s USB--> ethernet adapter a year or 2 ago, similar results:
[08:48:34] <pcw_home> it would run cheerfully for a hour or so and then get a 10 ms delay for no apparent reason
[08:50:22] <cradek> tinkerer: I'm unclear about what you're asking - are you offering to do this work, looking for collaborators, or asking for someone to do it?
[09:05:00] <tinkerer> cradek: @ 1st I'm checking if there some interests because in the past there were some resentments against lcnc on raspi.
[09:09:34] <archivist> I thought it was practical issues, speed and memory dunno how the new one shapes up
[09:14:04] <tinkerer> nice sweeping statement...
[09:19:20] <archivist> not intended as a sweeping statement at all, I dont have one just watch what people say
[09:19:44] <tinkerer> http://tom-itx.no-ip.biz:81/~tom-itx/irc/logs/%23linuxcnc-devel/2016-02-03.html#10:34:05
[09:32:52] <cradek> tinkerer: I bet some people are interested in it and others aren't, and it's OK to do work that not *everyone* agrees is worthwhile
[09:33:18] <cradek> and ... it's not a tablet, is it? I don't see how skunkworks's comment even applies
[09:35:54] <tinkerer> it was a sweeping statement in order to stop the discussion
[09:36:54] <tinkerer> but never mind
[09:39:39] <tinkerer> I need some reviewers for my sources due to the hacks I've made for compiling without a full source tree.
[09:39:39] <archivist> I did not get that impression at all
[09:40:19] <tinkerer> but particular? ;)
[09:40:33] <cradek> tinkerer: are you talking about adding a hardware driver to linuxcnc? we're absolutely interested in people doing that
[09:40:49] <cradek> tinkerer: please go on about what specifically you are asking for people to do
[09:45:22] <tinkerer> in the next few days I will push some code to my github repo then I will ask again.
[09:47:31] <cradek> cool
[09:47:54] <cradek> forking linuxcnc and putting your new stuff on a branch is a great way to share for review
[09:48:56] <tinkerer> this is the 2nd step
[09:49:46] <tinkerer> @ 1st I need a working env for lcnc on the raspi
[09:50:56] <cradek> is there a rt-preempt kernel for it? if so, I think linuxcnc will just build and run
[09:51:30] <cradek> because linuxcnc is known to build and run on armhf platform
[09:51:36] <tinkerer> otherwise it's too pingpong like :)
[09:52:22] <tinkerer> follow my link above :D
[09:52:42] <cradek> I did
[09:54:33] <cradek> what do you want me to look for there?
[09:54:59] <tinkerer> not now
[09:55:25] <cradek> (?)
[09:55:31] <cradek> I don't understand what you're saying
[09:55:58] <tinkerer> don't be to serious :)
[09:57:19] <cradek> I'm trying to help you
[09:57:37] <cradek> I'm often really unsure what you are asking/doing/wanting
[09:58:01] <tinkerer> it was a side blow about the discussion and my question in the link
[09:58:17] <cradek> what question?
[10:04:09] <tinkerer> arm support and cross compiling
[10:05:34] <cradek> building and running on the armhf platform IS known to work
[10:05:44] <archivist> and welcomed
[10:05:47] <cradek> I don't think our build system currently supports cross-compiling
[10:05:50] <tinkerer> yes but anyway
[10:08:22] <tinkerer> yes but this discussion leads to nowhere
[10:08:45] <tinkerer> now I will made my own image
[10:10:24] <tinkerer> but I will need some guidance
[10:10:42] <tinkerer> but not now
[10:11:14] <tinkerer> I will be back :)
[10:26:46] <jepler> pcw_home: I thought the outcome might be different with now that linuxcnc handles packet loss, but my experience is otherwise
[10:27:40] <jepler> It is not that we don't want to support armhf platforms or add hardware drivers. But the form in which the people now behind machinekit offered their code was not acceptable to us, which led to us going separate ways.
[10:28:15] <jepler> and somehow this has become ar umor that we don't want to accept code to run on trhese boards, which is false.
[10:28:20] <jepler> become a rumor
[10:28:50] <pcw_home> it may be dongle specific, I never got linkup/down problems just random dropouts of 8-10 ms
[10:30:36] <pcw_home> (with some random 1Gbe USB adapter)
[10:31:55] <tinkerer> it sounds like people behind a cross or a half moon....
[10:32:27] <tinkerer> I'm not behind mk. I just use it.
[10:32:47] <tinkerer> like lcnc
[12:30:34] <KGB-linuxcnc> 03Dewey Garrett 05joints_axes14 ca2f492 06linuxcnc 10src/emc/motion/control.c motion/control.c disable teleop_tp at softlimit JA * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ca2f492
[13:38:41] <seb_kuzminsky> http://devopsreactions.tumblr.com/post/144645169838/looking-at-old-code
[13:39:36] <cradek> roofers should get hazard pay
[14:20:52] <cradek> http://devopsreactions.tumblr.com/post/124823725833/systemd-evolving
[15:26:17] <Tom_itx> it's a wonder we survive at all: http://tom-itx.no-ip.biz:81/~webpage/women_vs_men/whymendie_index.php
[17:05:42] <jepler> (also I propose that working on CNC software could be nicknamed DevShops)
[17:12:40] <tinkerer> Q: "or specify --disable-gtk to skip the parts of LinuxCNC that depend on GTK"
[17:12:56] <tinkerer> which parts are this?
[17:47:22] <jepler> I would be very surprised if that flag works, nobody exercises it. it would control very old gtk-based parts of the system like halscope, if it works at all.
[17:57:08] <tinkerer> it sounds like "looking-at-old-code" :)
[17:58:04] <tinkerer> another Q: "Required OpenGL header missing." which package is it?
[18:00:50] <tinkerer> "Install it, or specify --disable-python to skip the parts of LinuxCNC that depend on Python" :D
[18:01:12] <tinkerer> sounds like "or don't use linuxcnc"
[18:10:12] <tinkerer> wow! i got the famous hashtag rectangle! :)
[18:24:10] <skunkworks> yay!
[18:24:44] <skunkworks> this laptop seems to run rt_preempt well
[18:24:58] <skunkworks> if you turn off speed step in the bios
[18:25:09] <skunkworks> otherwise it gets a bit hot
[18:30:06] <skunkworks> I had the previous rt_preempt kernel (4.4.7) cause reboots every so often. The latest doesn't seem to. so far (I would come back to the laptop in the morning and it would be at the login screen)
[18:48:54] <skunkworks> so far - this one hasn't rebooted
[18:49:34] <skunkworks> hasn't hmm - spell checking says hasn't is wrong.
[18:58:03] <seb_kuzminsky> tinkerer: GL/gl.h is in mesa-common-dev, and GL/glu.h is in libglu1-mesa-dev (on my Debian Jessie system)
[19:00:57] <tinkerer> seb_kuzminsky: thanks, but in the meantime it compiles :)
[19:02:53] <tinkerer> err. is compiling
[19:03:04] <seb_kuzminsky> welcome, happy to help
[19:04:11] <tinkerer> how long needs your arm buildbot for a fresh full make?
[19:05:20] <seb_kuzminsky> looks like ~33 minutes for full build & test
[19:05:21] <seb_kuzminsky> http://buildbot.linuxcnc.org/buildbot/builders/1405.rip-wheezy-armhf/builds/2341
[19:11:18] <tinkerer> heh, ok 58 min with pseudo cross compiling in qemu.
[19:11:47] <seb_kuzminsky> i tried qemu for arm builds (on a pretty fast amd64 host) and i was unhappy with the speed
[19:12:05] <seb_kuzminsky> i'm currently using an Odroid U3 for armhf builds
[19:12:17] <seb_kuzminsky> it's annoying in other ways, but it's relatively cheap and relatively fast
[19:12:29] <tinkerer> how can I make deb packages?
[19:13:08] <seb_kuzminsky> the simple way is:
[19:13:11] <seb_kuzminsky> cd debian
[19:13:17] <seb_kuzminsky> ./configure uspace
[19:13:18] <seb_kuzminsky> cd ..
[19:13:36] <seb_kuzminsky> dpkg-buildpackage -b -uc
[19:14:04] <seb_kuzminsky> that'll compile everything, build the debs, and deposit them in ..
[19:14:45] <seb_kuzminsky> the slower but more reliable way is to do the build in a controlled chroot, using pbuilder or sbuild
[19:14:48] <seb_kuzminsky> the buildbot uses pbuilder
[19:16:26] <tinkerer> If I have a finished compilation and I use the simple way does it recompile all?
[19:17:09] <seb_kuzminsky> you can look at debian/rules to see what it will do
[19:17:45] <seb_kuzminsky> you can see that it runs configure unconditionally, i think that will touch config.h and cause a bunch of recompilation
[19:18:03] <seb_kuzminsky> but i'm not totally sure - you'll have to try it
[19:18:18] <tinkerer> and whats your experience?
[19:18:23] <tinkerer> ok :)
[19:18:30] <seb_kuzminsky> i never do that, so i have no experience with it
[19:18:44] <seb_kuzminsky> i build debs from a clean tree, in order to make it as repeatable as possible
[19:19:08] <tinkerer> ok
[19:19:56] <tinkerer> unknown distribution: Raspbian-8.0
[19:20:20] <tinkerer> :) but successfully configured for 'uspace-Raspbian-8.0'-'uspace'..
[19:20:39] <seb_kuzminsky> yeah, debian/configure tries its best when you run it on some distro it doesn't know about
[19:20:56] <seb_kuzminsky> bbl, time for dinner
[19:20:59] <seb_kuzminsky> good luck!
[19:21:39] <tinkerer> bon appetit
[19:21:45] <tinkerer> thanks
[19:39:18] <tinkerer> ok, I copied the debian-8.*) case and renamed it Raspbian-8.*)
[19:43:25] <tinkerer> wow, what a mess...
[19:50:33] <jepler> oh I forget, even though pi2/3 can run unmodified debian armhf userspace, "raspbian" still has a lot of traction and an incompatible ABI. That's unfortunate.
[19:52:15] <tinkerer> a mess of squares?
[19:55:36] <tinkerer> uebasht
[19:55:51] <tinkerer> oops wrong window
[20:32:10] <jepler> time to change your password :-/