Back
[06:56:49] <archivist> I presume PNCconf uses its own copy and trashes hal and ini files like stepconf ?
[07:02:58] <jthornton> aye I think so
[07:06:51] <archivist> I think that may also be confusing person next door :(
[07:16:11] <jthornton> yes for sure
[07:17:27] <archivist> I had a quick look at the docs and that bug is less prominent than in stepconf docs
[14:48:57] <skunkworks> http://www.cnczone.com/forums/linuxcnc-formerly-emc2-/299032-linuxcnc-2-7-3-g0-z0.html
[15:08:43] <KGB-linuxcnc> 03chris morley 052.7 321c58b 06linuxcnc 10src/emc/usr_intf/pncconf/private_data.py pncconf: fix wrong inverted step/direction pin * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=321c58b
[15:22:53] <seb_kuzminsky> skunkworks: maybe GeeB is seeing a commanded-vs-actual position display difference? he's only off by 0.1 mm, maybe that's smaller than the step size on his machine, or the tuning on his servos?
[15:30:03] <skunkworks> seb_kuzminsky: good question. also - is he seeing it in the dro - or is it an offset he didn't know about
[15:30:09] <skunkworks> (by measuring)
[15:33:49] <seb_kuzminsky> skunkworks: good question, i was assuming he was reading the DRO but he never said that
[15:35:17] <skunkworks> seb_kuzminsky: so if I understand it right - the new kernel has rt_preempt integrated into it?
[15:48:14] <jepler> skunkworks: it's still a separate patch
[15:48:22] <jepler> .. but the debian people have put that patch in their kernel as an option again
[16:01:02] <seb_kuzminsky> yep, that's it
[16:01:25] <seb_kuzminsky> it's not been pushed to sid's repo yet, i'm building it locally right now, on jessie
[16:09:07] <skunkworks> awesome
[16:09:16] <skunkworks> sid is debian unstable?
[16:15:19] <jepler> yes
https://www.debian.org/releases/sid/
[16:36:57] <jepler> well this is revolting
[16:37:12] <jepler> linux-image-4.4.0-trunk-rt-amd64 breaks xserver-xorg-input-vmmouse (<< 1:13.0.99)
[16:37:14] <jepler> xserver-xorg-input-vmmouse (version 1:13.0.0-1+b3) is present and installed.
[16:37:47] <jepler> .. but xserver-xorg-input-vmmouse can't be removed without removing task-xfce-desktop
[16:37:58] <seb_kuzminsky> dpkg --remote --force-depends
[16:38:02] <seb_kuzminsky> *--remove
[16:38:13] <jepler> yeah but you can't tell humans to do that
[16:38:27] <seb_kuzminsky> you're probably right
[16:38:55] <seb_kuzminsky> i'm finding 4.4.1-1_exp1 won't build on jessie
[16:39:16] <jepler> I banged it until it used gcc-4.9 on i386 and amd64
[16:39:28] <seb_kuzminsky> it needs a newer pbuilder, and it build-depende on gcc-5, which isnt in jessie
[16:39:37] <jepler> newer pbuilder?!
[16:39:55] <seb_kuzminsky> the jessie pbuilder doesnt handle the new kernel BuildProfiles syntax
[16:40:07] <seb_kuzminsky> the pbuilder from stretch works, and installs without problems on jessie
[16:41:02] <jepler> I see.
[16:41:52] <jepler> nopaste
[16:42:03] <jepler> App::Nopaste::Service::Codepeek: Error GETing
https://www.HugeDomains.com/shopping_cart.cfm?d=CodePeek&e=com&paste=diff+-Nru+linux-4.4.1%...
[16:42:06] <jepler> uhhh
[16:42:14] <jepler> wth is nopaste doing
[16:42:32] <seb_kuzminsky> paying someone $2500, looks like
[16:43:09] <jepler> oh codepeek.com is now domain squatted
[16:43:32] <jepler> anyway I think that this is the bit to use gcc 4.9:
http://paste.debian.net/387202/
[16:43:54] <jepler> but the whole debdiff is big because it includes a lot of files that get rebuilt (generated files in debian/)
[16:46:42] <seb_kuzminsky> jepler: that looks similar to my diff ;-)
[16:48:41] <jepler> a 1-minute latency of 19.9us (latency-histogram --nobase --sbinsize 1000)
[16:49:08] <jepler> no latency spike from glxgears, whether running on intel or nvidia video (it's a dual-video laptop)
[16:49:15] <Tom_itx> off the wall question but did any of you ever code in ca-clipper way back when?
[16:49:19] <jepler> oops I spoke too soon, something gave me 300us latency
[16:49:47] <Tom_itx> trying to figure out their multidimensional array access syntax
[16:49:58] <jepler> Tom_itx: doesn't ring any bells
[16:50:03] <Tom_itx> ok
[16:50:08] <Tom_itx> was worth a shot..
[16:52:33] <jepler> (I don't recall what I got with the 4.1.x I was using for the longest time either)
[16:59:41] <skunkworks> 3.14? (wheezy repository) gives me about 80us - peter says new versions are alot better
[17:07:15] <seb_kuzminsky> skunkworks: i'll have a deb for you to test tonight ;) 4.4.1-rt5
[17:12:42] <skunkworks> ooh
[17:49:43] <KGB-linuxcnc> 03chris morley 05master 0ae98da 06linuxcnc Merge branch '2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=0ae98da