#linuxcnc-devel | Logs for 2017-01-27

[12:40:39] <skunkworks_> zlog
[13:05:53] <skunkworks_> pcw_mesa, A bit of a difference - seems -100 is the best.. (vs -50) If I did that right
[13:06:16] <skunkworks_> http://electronicsam.com/images/KandT/testing/emco/predpll.png
[13:06:31] <skunkworks_> http://electronicsam.com/images/KandT/testing/emco/postdpll.png
[13:20:48] <pcw_mesa> Would be more noticeable if the host latency was bad (but only bad when you have latency spikes)
[13:22:47] <pcw_mesa> Its very noticeable with the stepgen since the PID-Stepgen is an almost perfect velocity mode control model
[14:24:08] <JT-Shop-2> pcw_mesa: do you have the instructions for building rt preempt on mint 18?
[14:24:15] <JT-Shop-2> I had them but lost them :)
[14:24:17] <JT-Shop-2> :(
[14:24:47] <pcw_mesa> I dont think they are any different than for Mint17
[14:25:15] <JT-Shop-2> ok, what kernel are you building for 18?
[14:25:34] <pcw_mesa> there are still a few LinuxCNC bugs with Mint18 the probably need to be "Jeplered"
[14:26:42] <JT-Shop-2> ok I'll wait
[14:27:18] <pcw_mesa> I dont think the kernel version is terribly important other than Ive found the kernel newer than 4.1.x tend to have bad latency on older hardware (like my Core Duos)
[14:29:41] <pcw_mesa> I have a mint 18 system running 24 7 here for a couple Months using 4.1.35-rt31 (Its a HP Elite 8300 I5 3470 USDT)
[14:32:40] <pcw_mesa> also same thing running on a a Core duo (HP DC7900 USDT) with LinuxCNC up about 6 Months with a Ethernet I/O card
[14:32:40] <JT-Shop-2> thanks, I'll give that a whirl in the morning
[14:34:03] <pcw_mesa> Very new hardware (like intel N3150, Z8500s, Skylake etc) will need a newer kernel
[15:40:02] <stustev> what is the purpose (meaning) of convert in the interp_convert.cc? What is the purpose of converting?
[17:35:02] <andypugh> Help! :-) How do we get the command-line latency test now?
[17:35:25] <andypugh> The one that lets you spot a periodic problem
[17:40:11] <stustev> cd to linuxcnc/scripts and type ./linuxcnc-tests
[17:40:49] <stustev> if you have modified your .bashrc file you should be able to type the command whereever you are
[17:43:13] <andypugh> I don’t think that’s the one I was after
[17:50:23] <stustev> I guess I don't know of any other
[18:00:09] <seb_kuzminsky> andypugh: scripts/latency-test?
[18:01:24] <andypugh> Pops up the GUI, I think
[18:02:08] <andypugh> There was one that updated in the terminal every second, as I recall, from whan I cared, about 8 years ago
[18:02:32] <seb_kuzminsky> right, i think you're right
[18:03:38] <andypugh> latency-plot might be the new choice
[18:03:44] <seb_kuzminsky> in rtai-land, there's /usr/realtime*/testsuite/kern/latency
[18:04:23] <seb_kuzminsky> latency-plot is graphical too, no?
[18:05:21] <seb_kuzminsky> i like latency-histogram, but it too needs a gui
[18:05:50] <kwallace2> stustev, my understanding of interp_convert.cc is that it takes in the parsed g-code and converts it to NML machine commands, such as "STRAIGHT_TRAVERSE(block->line_number, end_x, end_y, end_z, ... ", as well as doing some error checking and notification.
[18:06:26] <seb_kuzminsky> kwallace2: that's my interpretation (heh) of the name, too
[18:10:29] <JT-Shop> seb_kuzminsky: I wrote a gui for the latency histogram
[18:10:30] <andypugh> kwallace1: I think that STRAIGHT_TRAVERSE (and freinds) are actually canononical motion commands, rather than NML command.
[18:10:37] <JT-Shop> lets you select the options
[18:11:24] <andypugh> Though they are canonical motion commands encoded in NML mainly to make things harder to follow :-)
[18:12:16] <stustev> kwallace2: thanks a lot - that makes it easier to read
[18:13:33] <stustev> andypugh: you are correct - maybe a little verbage (is that like verbal garbage?) in the file would help
[18:15:48] <kwallace2> my exposure to interp_convert and interp_check was to modify G76 to do tapered threads but I didn't go much beyond that. There seems to be a lot of feature overlap in these and related files.
[18:17:12] <andypugh> Yeah, we should just start again from scratch. :-)
[18:19:54] <kwallace2> It might be interesting to try to go back and try to load and run EMC1.
[18:21:05] <andypugh> All the overhead was already in by then. HAL came later, but HAL is actualy the best bit.
[18:28:03] <andypugh> Are stepconf and pncconf JA-aware? This thread made me wonder, though beither are mentioned by name. https://forum.linuxcnc.org/gmoccapy/32227-bug-report-gmocappy-homing-buttons-in-joint-mode-in-master#86682
[18:34:53] <andypugh> Also prompted by a forum post: Halshow (Show HAL Config) Why is Red == 0 and Yellow == 1 ?
[18:59:22] <pcw_mesa> related to the pathpilot post where the user was colorblind?
[18:59:44] <pcw_mesa> (so changed the LEDs to black/green)
[19:02:30] <pcw_mesa> (standard Pathpilot LED colors are Yellow/Green)
[19:09:38] <andypugh> Black/White. Why not?
[19:28:58] <kwallace2> Using graphics to communicate status doesn't seem ideal to me. I tend to like words. "Power is: On" "Power is: Off". I started using an Android pad and I can't figure out what symbol I need to send an e-mail message to save my life.
[20:05:22] <pcw_mesa> I think I would like black/green instead of LinuxCNCs current red/yellow
[20:41:30] <andypugh> I agree, and that seems like it should be an easy fix, and nit ambuguous (both colours change)
[20:42:58] <andypugh> Res for false and yellow for true seems a truly bizzare choice, so I assume there is a reasn it has persisted for 10 years>
[21:03:29] <KGB-linuxcnc> 03Dewey Garrett 05master d3a8b92 06linuxcnc 10configs/sim/axis/gantry.ini gantry.ini: use neg homing_sequence for y joints * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=d3a8b92
[21:27:03] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/external_offsets 98b9d98 06linuxcnc 10(16 files in 6 dirs) external axis-offset hal pins (v11) * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=98b9d98
[21:44:50] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/external_offsets 060fa4d 06linuxcnc 10(17 files in 7 dirs) external axis-offset hal pins (v12) * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=060fa4d
[22:34:32] <andypugh> why does EBo make every simple question more complicated?
[22:48:40] <linuxcnc-build_> build #4712 of 1306.rip-precise-amd64 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1306.rip-precise-amd64/builds/4712 blamelist: dummy, Dewey Garrett <dgarrett@panix.com>
[23:38:53] <linuxcnc-build_> build #4727 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4727 blamelist: dummy, Dewey Garrett <dgarrett@panix.com>