#linuxcnc-devel | Logs for 2014-10-16

[10:28:54] <seb_kuzminsky> am i the only one who often wishes for our gcode overview in manpage format?
[10:29:21] <ssi> never thought about that, but now that you mention it that'd be handier'n pants pockets :D
[10:29:43] <seb_kuzminsky> heh, that was the response i was hpoing for, i'll eagerly await your PR now ;-)
[10:34:43] <ssi> hahaha
[10:40:38] <cradek> man g81
[10:40:42] <cradek> ?
[10:48:04] <cradek> mozmck: are you coming to houston?
[10:48:46] <mozmck> hi cradek: unfortunately I don't think so.
[10:48:58] <cradek> darnit, I was hoping we could jam.
[10:50:12] <mozmck> yeah, that would be fun!
[10:50:40] <mozmck> wife is almost 8 months pregnant now, and too much else going on.
[10:51:15] <cradek> aha, that's more important than travels; hope everything comes out ok
[10:52:05] <mozmck> thanks!
[11:00:03] <cradek> while packing, I remembered the joke about the guy who left his banjo in his unlocked car
[11:15:18] <pcw_home> I guess I will need to have a network setup script for Ethernet cards or people will fight with the network mangler forever
[11:56:20] <pcw_home> Damn, the sheep have figured out how to get through chicken wire...
[12:01:39] <seb_kuzminsky> cradek: i was imagining "man gcode" showing basically the G-code Quick Reference, http://linuxcnc.org/docs/html/gcode.html
[12:02:05] <seb_kuzminsky> just enough to remind me how to spell "absolute arc centers" and stuff like that
[12:19:49] <jepler> pcw_home: following the hm2_eth manpage section "configure network with standard address" should stop network-manager touching the dedicated ethernet interface after the next reboot
[12:21:46] <jepler> if n-m is not hands off after making that configuration, then I need to figure out what to write in the docs
[12:23:04] <pcw_home> I think that works but if you poke an the interface with NM ist still breaks things (so dont do that I guess)
[12:27:43] <seb_kuzminsky> pcw_home: would this help? http://www.campmor.com/72-tight-weave-no-see-um-mosquito-netting.shtml?source=CI&ci_sku=61001&ci_gpa=pla&ci_kw=
[12:27:57] * seb_kuzminsky <-- so helpful
[12:28:13] <pcw_home> not sure if setting it up entirely outside of NM and deleting any of NM connection profiles that refer to the RT ethernet connection woudl make it invisable to NM
[12:32:00] <pcw_home> I didnt think the sheep could get through the chicken wire that overlays the welded wire (to prevent nose incursion)
[12:32:02] <pcw_home> that protects our small trees from sheep depradation but holes are appearing in several places...
[12:33:10] <pcw_home> Probably underestimated the effect of small sharp hooves and determination
[12:35:48] <archivist> I am having a similar problem with starlings, three days on the trot, had to get one out of the house, hole in the eaves somewhere
[12:40:06] <pcw_home> Problem with the sheep is they have pretty much 24 hours a day to work on getting into things
[12:44:00] <seb_kuzminsky> you should give them some thing productive to occupy their time
[12:44:02] <seb_kuzminsky> pick-n-place maybe
[12:45:49] <pcw_home> Rather difficult to motivate except with food
[12:47:50] <archivist> when I caught the starling this morning it grabbed a cable and would not let go. fun uncurling its claws to throw it out
[13:21:58] <pcw_home> Didn't know they were that tenacious
[13:29:55] <CaptHindsight> growing sweaters and socks vs all that shearing and knitting, cut out the middle man
[14:51:51] <PCW> interestingly enough, Jitter on Ethernet connected cards under Preemt-RT is not substantially worse than jitter on a D525 under RTAI (~120 usec read time jitter)
[14:53:34] <PCW> (random delay spikes in read function 120 usec longer than ~30 usec baseline)
[14:55:50] <PCW> these spikes affect all functions, even those that do not access hardware
[15:13:04] <jepler> $ echo 1 | sudo tee /proc/irq/45/smp_affinity
[15:13:05] <jepler> 1
[15:13:14] <jepler> apparently this is the right way to set an irq's smp affinity
[15:13:31] <jepler> I had mistakenly thought that it was effective to change the affinity of the task seen in 'ps' with cpuset
[15:13:59] <jepler> in my system, 45's eth0 and I could see from /proc/interrupts that it was being delivered to the RT CPU
[15:15:11] <jepler> NMIs, local timer interrupts, and performance monitoring interrupts are also delivered on my RT CPU
[15:16:26] <jepler> possibly it's also beneficial to set the eth1 interrupt's affinity *to* the RT CPU
[15:17:48] <jepler> hm I'm booted without isolcpus=
[15:21:22] <PCW> I tried messing with irq affinities a while back and did not see any noticeable change but its very likely I messed something up
[15:22:37] <jepler> looks fiddly to get right
[15:22:47] <jepler> and I'm not in a position right now to measure whether it's making an effective difference
[15:23:33] <PCW> Yeah the thing I dont get is these 100 usec latencies under RTAI
[15:24:40] <CaptHindsight> have you tried the new RTAI tree?
[15:25:10] <PCW> No, how much trouble is it to try?
[15:25:14] <CaptHindsight> I'm wondering if it behaves any different on Intel
[15:25:35] <CaptHindsight> making deb kernels are easy
[15:27:19] <CaptHindsight> memleak fiddled quire a bit with the scheduler to keep the latency on the RTAI core low while having smooth video behavior
[15:27:47] <CaptHindsight> I wonder if your spikes are to do with video
[15:28:21] <PCW> have you plotted say motion-controller.time vs video access (and motion-controller.tmax)
[15:28:54] <PCW> halscope it and move the mouse :-)
[15:29:24] <CaptHindsight> he did all the work, so I don't know
[15:29:37] <PCW> or watch tmax tic upwards if you play a youtube video
[15:32:11] <CaptHindsight> PCW: does it creep up the longer it plays for you?
[15:32:44] <CaptHindsight> we'd run it for 1-2 weeks and never see a spike
[16:01:53] <CaptHindsight> the trouble with RTAI is that nobody really knows how it all works and why at this point
[16:02:57] <seb_kuzminsky> i agree, that's very troubling
[16:03:32] <seb_kuzminsky> a while back jepler (i think) suggested making an rtapi module for i-pipe, which seems appealing to me
[16:04:01] <seb_kuzminsky> i-pipe is much of the awesome sauce of rtai, and it seem much more alive than rtai
[16:07:20] <CaptHindsight> memleak put everything in his own repo since Paolo hardly updates anything and is currently borrowing from xenomai
[16:08:57] <CaptHindsight> http://github.com/NTULINUX/RTAI
[16:10:36] <CaptHindsight> I don't who the main users of RTAI are anymore besides this project
[16:49:19] <seb_kuzminsky> yeah memleak is probably as productive as the rest of the rtai community put together
[17:20:21] <jepler> http://www.troll.me/images/cowboy-dos-equis/i-dont-always-nullterminate-my-strings-but-when-i-do-it-shows.jpg
[17:22:02] <jepler> strlcpy / strncpy / strzcpy / strcpy_s
[17:22:10] <jepler> it's 2014 and we're still not sure of the best way to copy "strings" in C
[17:23:35] <seb_kuzminsky> i'm not gonna click on "troll.me"
[17:23:41] * seb_kuzminsky clicks anyway
[17:26:32] <jepler> seb_kuzminsky: plans for houston still on the rails?
[17:27:53] <jepler> seb_kuzminsky: fwiw my flight gets in at ~8:30pm friday to iah and I don't know if I'm going to try to do anything friday night
[17:29:55] <cradek> I'm having dinner in wichita already
[17:30:20] <jepler> cradek: hi wichita
[17:31:27] <cradek> seb_kuzminsky: what's your flight? You need a pickup, right?
[17:33:38] <jepler> cradek: you didn't miss much at $DAY_JOB trade show today. software still looks terrible on a projector, and every new feature is demonstrated with an apology at the start like always
[17:35:37] <seb_kuzminsky> that's depressing
[17:36:32] <seb_kuzminsky> i get in to iah at 12:05 tomorrow, god willin and the crick dont rise
[17:36:36] <jepler> seb_kuzminsky: when asked, our talk givers say that they are trying to say to the users: this is different than it has been in the past, and at first you'll be irritated by the difference itself, but trust us it's better
[17:36:44] <jepler> at least that's what they say to us
[17:37:09] <seb_kuzminsky> jepler: ah, that's much better than what i was imagining
[17:37:36] <seb_kuzminsky> cradek: i'd like a ride if it's convenient, or i'll hire a cab, no big deal either way
[17:37:43] <seb_kuzminsky> (ie, don't speed to get there to pick me up)
[17:38:01] <jepler> or you can wait for 8PM to roll around
[17:38:25] <cradek> I doubt I'll be there anytime near noon but you never know. coffee with dinner.
[17:40:09] <seb_kuzminsky> cradek: i'll probably just meet you at the hotel then
[17:40:14] <jepler> cradek: I'll give you a call when I get in, find out if anything's up
[17:40:50] <seb_kuzminsky> by 8 in the pm we should be well into the "discuss moonshots over beer" part of the meeting? or hacking feverishly on our giant backlog of bugs, one of those two probably
[17:41:10] <jepler> what giant backlog of bugs? I looked on github linuxcnc-mirror and there are zero bugs
[17:41:24] <seb_kuzminsky> heh
[17:42:03] <jepler> great, google knew before I finished typing it in the specific beer I'm having
[17:43:39] <jepler> hmm actually this description does not sound at all like the beer I'm having
[17:44:17] <jepler> aha, google was wrong. I'm having the 18. http://www.beeradvocate.com/beer/profile/392/95268/
[17:44:32] <jepler> it's good as long as you like a malty, high alcohol beer with nary a hint of hops
[17:45:40] <seb_kuzminsky> TIL jeff and i are in opposite quadrants of the hops-vs-malt chart
[17:45:54] <jepler> seb_kuzminsky: I thought we knew that already
[17:46:01] <jepler> at least I remember barely being able to manage a sip of the "F Y IPA"
[17:46:06] <jepler> in your presence
[17:47:48] <seb_kuzminsky> heh oh yeah
[17:47:53] <seb_kuzminsky> wow that seems like a long time ago
[17:48:18] <jepler> it was probably 5-7 years
[17:48:59] <jepler> man I am looking forward to hanging out
[17:49:08] <seb_kuzminsky> :-)
[17:49:26] <seb_kuzminsky> me too, and speaking of which imma ditch out of work early to go pack now :-)
[17:49:33] <seb_kuzminsky> see some subset of you all soon!
[21:31:37] <skunkworks> logger[psha]:
[21:31:37] <logger[psha]> skunkworks: Log stored at http://psha.org.ru/irc/%23linuxcnc-devel/2014-10-17.html
[21:47:49] <jepler> wow, I'd love a 5120 x 2880 (27" ~220dpi) display. too bad there's a mac attached to it.
[22:03:19] <ssi> dell makes one too
[22:03:25] <ssi> too bad it costs as much as one with a mac attached to it
[22:07:55] <skunkworks> have a good trip!
[22:09:06] <ssi> I wish I was going :/
[22:09:12] <ssi> I would if I hadn't spent all my dollars on the vmc :P
[22:17:29] <skunkworks> heh - we are just too busy. would have loved to go
[22:24:44] <cradek> yay, made it almost to texas.
[22:40:37] <Tom_itx> stopped for the night?
[22:42:42] <cradek> yeah, when tail lights change from points to big circles, it's time to stop
[22:42:55] <cradek> it's weird but predictable
[22:43:25] <cradek> jeez texas is tall (5 hours to go tomorrow)
[22:46:22] <cradek> living smack in the middle of the US, I can get to any other spot in about 1.5 days
[22:46:48] <ssi> having an airplane., I can get to any other spot in the continental us within 1.5 days :D
[22:46:52] <skunkworks> cradek: did you see http://electronicsam.com/images/KandT/testing/Screenshot%20-%2010162014%20-%2007:12:14%20PM.png
[22:47:10] <skunkworks> some version of opencv installs on wheezy
[22:47:14] <cradek> no!
[22:47:20] <cradek> what is it?
[22:47:34] <cradek> I ordered one of those cameras -- too cheap not to
[22:47:41] <skunkworks> (shape recognition..)
[22:48:27] <skunkworks> I have a python program talking to hal from opencv.. looks doable..
[22:48:46] <cradek> neat!
[22:48:58] <cradek> what's your overall goal?
[22:49:32] <skunkworks> really just some applications for work.. (positioning stock by targets for cutting or printing..)
[22:49:43] <cradek> awesome
[22:49:52] <skunkworks> we have machines that do that now - but they are really really expensive.
[22:50:24] <cradek> you guys seem to replace really really expensive things with homemade things frequently
[22:50:31] <skunkworks> it is pretty cool - I didn't think it would be that easy.... Well - you know..
[22:50:45] <cradek> easy is sure relative
[22:50:49] <skunkworks> right
[22:51:10] <cradek> possible (free software building blocks) is great
[22:51:24] <skunkworks> yes - business is picking up and they are trying to build more machines.. getting an extra cnc maybe...
[22:51:48] <skunkworks> they have someone actually using the one they have now. (finally)
[22:52:51] <skunkworks> cradek: the camera is suprisingly good for $27
[22:53:12] <skunkworks> I showed it to a co-worker and he ordered one :)
[22:53:17] <cradek> manual focus from touching to infinity sounds really good
[22:53:28] <skunkworks> I wasn't expecting that..
[22:53:32] <cradek> "just works" with v4l is also really good
[22:53:57] <skunkworks> I have an old logitech camera that doesn't work..
[22:54:15] <skunkworks> I though - it is old - it should work great!.
[22:54:43] <cradek> I have a decent microsoft branded webcam that works
[22:54:54] <cradek> it focuses by itself at random, which I suppose is a feature
[22:55:15] <cradek> I meant to make an internet bird-watcher out of it, but that's stupid so I didn't
[22:55:43] <skunkworks> heh.. the company that sells the supereye also sells an autofocas one - $200. (according to my coworker that researches things to death)
[22:57:32] <cradek> funny
[22:57:41] <cradek> I bet manual is much better for things we're likely to do
[22:59:11] <skunkworks> right
[22:59:24] <skunkworks> 6 turns from one extreme to the other
[23:13:19] <skunkworks> hard to hold it steady
[23:13:21] <skunkworks> http://electronicsam.com/images/KandT/testing/2014-10-16-225003.jpg
[23:13:41] <skunkworks> http://electronicsam.com/images/KandT/testing/2014-10-16-225142.jpg
[23:13:48] <skunkworks> (both by the same camera)
[23:15:29] <cradek> wow
[23:15:56] <skunkworks> a bit of backlash in the focus - but usable.
[23:16:01] <cradek> actually looks useful as a microscope
[23:16:15] <skunkworks> yes
[23:16:34] <skunkworks> (it has to be all but touching to get that close...)
[23:17:55] <skunkworks> but it has led's in the tube (which you can adjust the intensity) so you can touch it without having to worry about ligh
[23:17:57] <skunkworks> light
[23:18:29] <cradek> that makes it *really* useful as a microscope
[23:21:55] <skunkworks> what - you didn't like the mirror you had to adjust 'just right' to get be able to see anything in the old microscopes?
[23:21:57] <skunkworks> ;)
[23:23:20] <cradek> my current one has both a top (top and back) and bottom light, but sometimes neither works if your hands are in there too
[23:23:45] <cradek> having the light come from where you're looking through is ideal
[23:24:09] <cradek> but of course I'm not going to throw out my stereo microscope anytime soon
[23:25:21] <skunkworks> heh
[23:26:00] <skunkworks> hmm - mach has a new 'tangential planner' now... does that mean 'circular arc blending'?
[23:27:08] <cradek> hm, I should try to sleep
[23:27:24] <skunkworks> isn't it midnight your time?
[23:27:32] <cradek> 11 central
[23:27:45] <cradek> I think I'm still in central here, but who knows for sure
[23:28:04] <skunkworks> oh - I thought for some reason you where mountain time..
[23:28:12] <cradek> it's 10 mountain
[23:28:18] <skunkworks> heh
[23:28:25] <skunkworks> maybe I need to sleep...
[23:28:30] <cradek> yeah, sounds like you need to go to bed too
[23:28:45] <cradek> and to all a good night!
[23:28:48] <skunkworks> have fun!
[23:31:50] <cradek> hmm, chris k says he wants someone to give a talk about servo tuning
[23:32:10] <cradek> oh good, jon is going to do it
[23:32:24] <cradek> goodnight