#linuxcnc-devel | Logs for 2016-08-28

[14:53:54] <cradek> kwallace2: I've heard about that a couple times - a new dependence in 2.7 between F word and threading
[14:54:01] <cradek> if you can nail it down, please make a bug report
[14:57:33] <kwallace2> In looking at G76 more, the current G76 doesn't call a feed move that is not tied to synchronized motion, so it should not be a problem. In one of my versions I had an initial in-feed because I wasn't comfortable with a rapid towards the workpiece.
[14:58:02] <kwallace2> My version that does rapid, works fine without an F value.
[14:58:49] <kwallace2> I guess I need to remember to check for F when I need it.
[14:59:29] <cradek> yeah I suppose if you send a feed you had better have previously checked for the F word, like G1 does
[15:04:02] <kwallace2> I took me a long time to figure out, sort of like adding a component and forgetting to assign it to a thread.
[15:09:35] <cradek> what does motion do if it gets an F=0 feed? seems like it should give an error...
[15:13:43] <kwallace2> Here was my edit: http://wallacecompany.com/tmp/Screenshot-g76_sync_prob-2a.png
[15:14:08] <kwallace2> This is what it did: http://wallacecompany.com/tmp/Screenshot-g76_sync_prob-1a.png
[15:16:46] <kwallace2> The straight feed was skipped, then the synch invoked and the next feed proceeded.
[15:18:28] <kwallace2> Oops, the first traverse on the right side should be a feed.
[15:20:52] <kwallace2> Oops, again, the right side is accurate.
[16:27:09] <skunkworks> zlog
[16:28:12] <skunkworks> jessie and linuxcnc running rt_preempt on the matsuura
[16:45:52] <seb_kuzminsky> skunkworks: nice
[16:46:04] <seb_kuzminsky> did you find the buildbot debs or build your own?
[16:52:58] <skunkworks> I ended up building the 4.6.7-rt11
[16:53:27] <skunkworks> letting it run over night with linuxcnc/mesa 7i80 running to see if there are any realtime delays
[16:56:54] <pcw_home> did you just update to Jessie?
[16:58:53] <skunkworks> yes
[16:59:11] <skunkworks> I didn't do any isolcpu or anything yet
[16:59:56] <pcw_home> pretty sure Isolcpus is not of much value on Preemt-RT
[17:00:27] <pcw_home> (or you need some other magic for it to help)