#linuxcnc-devel | Logs for 2016-08-22

[07:54:38] <skunkworks> zlog
[09:36:16] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15SebKuzminsky commented on issue #126: @VladAlpha, the behavior on program end (M2 or M30) is documented here: http://www.linuxcnc.org/docs/html/gcode/m-code.html#mcode:m2-m30... 02https://github.com/LinuxCNC/linuxcnc/issues/126#issuecomment-241424046
[10:18:28] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15zultron commented on issue #126: Here's what abort has always done, or anyway done since at least 2.6 (please check my statements):... 02https://github.com/LinuxCNC/linuxcnc/issues/126#issuecomment-241437936
[14:51:15] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15gmoccapy commented on issue #157: I just tested again with the most recent 2.7 branch, it is still wrong.... 02https://github.com/LinuxCNC/linuxcnc/issues/157#issuecomment-241521534
[14:53:33] <cradek> ok I tested 2.6 and he's testing 2.7
[14:55:06] <jepler> seb_kuzminsky: it's possible to underclock the u3, multiples of 100MHz down to 200MHz. Maybe underclocking it by a few hundred would help reliability?
[14:55:18] <jepler> /sys/devices/system/cpu/cou0/cpufreq/scaling_max_freq
[14:55:39] <jepler> default is apparently 1704000, other choices are all nice round numbers like 1200000
[14:56:16] <jepler> if you're using the on-board fan there are also tunables for the start temperature and duty cycle, though if I recall correctly you're pointing some other fan at it
[14:56:49] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15cradek commented on issue #157: I just tried the same tests I described above in v2.7.6-14-gdf183a4 and it also works how I expect, same as described above. I will wait for more information, thanks.... 02https://github.com/LinuxCNC/linuxcnc/issues/157#issuecomment-241522975
[15:04:40] <seb_kuzminsky> jepler: yeah, my fan died and i switched back to a passive heatsink, with some random salvaged case-fan sitting on the desk next to it
[15:04:56] <seb_kuzminsky> last night i ordered two more U3 heatsink+fan combos from Hardkernel
[15:05:04] <seb_kuzminsky> $10 in parts and $16 in shipping.... :-/
[15:06:51] <jepler> seb_kuzminsky: boo
[15:07:12] <jepler> I just got my XU4 today
[15:07:20] <seb_kuzminsky> ooh shiny
[15:07:31] <jepler> I hope I can get debian jessie and an RT kernel on it
[15:10:51] <jepler> one interesting problem it'll present: it is a so-called heterogeneous multi core, also called big.LITTLE. So the highest numbered CPU in /proc/cpuinfo turns out to be a less-powerful one, and that's the core linuxcnc will select
[15:11:46] <cradek> that's probably a fine choice, isn't it?
[15:11:52] <seb_kuzminsky> i was thinking that too
[15:12:09] <seb_kuzminsky> should be plenty for the hal components, and leave the big cpus for gremlin etc
[15:14:08] <jepler> maybe so
[15:14:20] <jepler> as long as you're not trying to have a fast base-period anyway
[15:14:34] <cradek> or run kins
[15:14:40] <jepler> embedded -> (spi) -> 7i90 is the idea that makes the most sense to me
[15:14:43] <jepler> cradek: oh good point
[15:14:55] <cradek> but you really need a real computer for that
[15:14:59] <jepler> though I really don't have a sense of how CPU hungry kins are, vs a mere 1.2GHz ARM CPU core
[15:15:16] <cradek> american robot was bad on a d525 or however that model was spelled
[15:15:18] <jepler> genhexkins more, lineardeltakins less
[15:15:23] <jepler> genserkins more
[15:15:26] <seb_kuzminsky> i bet embedded boards these days are faster than the under-the-desk tower PCs of yesteryear
[15:15:31] <jepler> I miss American Robot.
[15:15:44] <cradek> wonder where it ended up
[15:15:47] <jepler> I wonder if it's at Stuart's new shop, yeah
[15:15:54] <jepler> if not, .. midnight raid?
[15:16:10] <seb_kuzminsky> i'm planning to bring my little scorbot again, it's more portable though less cool
[15:16:12] <cradek> only if you promise to take it home, I bet
[15:16:25] <jepler> I'm debating whether to take my linear delta
[15:16:33] <seb_kuzminsky> because it's not strong enough to pick me up and rub me on the concrete floor like a piece of chalk
[15:17:00] <jepler> I'd have to buy stepper drivers and do wiring to make it move with linuxcnc and it's hard to think of that as fun
[15:17:16] <cradek> but we already did all that
[15:17:17] <jepler> it is .. big enough to be inconvenent to get into my car
[15:17:23] <cradek> now that's true
[15:17:38] <seb_kuzminsky> any idea how much it weighs?
[15:17:58] <seb_kuzminsky> is it "3 people can pick it up" or do you need a fork lift?
[15:18:03] <jepler> cradek: we wired it up to one of those dedicated grbl boards with integrated "overheat by design" heatsinkless stepper drivers, not an appropriate starting point for a linuxcnc conversion
[15:18:10] <jepler> seb_kuzminsky: oh one person can carry it around
[15:18:20] <cradek> wait
[15:18:23] <seb_kuzminsky> wait
[15:18:25] <cradek> we're talking about different things
[15:18:35] <seb_kuzminsky> the lineardelta3dprinter is 1-person
[15:18:36] <jepler> you're talking about scorbot
[15:18:43] <seb_kuzminsky> but what about the American Robot?
[15:18:44] <jepler> you're talking about american robot?
[15:18:50] <seb_kuzminsky> the scorbot is 1-person-portable
[15:18:58] <cradek> scorbot doesn't fit in your car!?
[15:19:09] <cradek> it's like 2 cubic feet isn't it?
[15:19:15] <jepler> delta < scorbot < american robot
[15:20:06] <Tom_itx> when is that meet?
[15:20:22] <Tom_itx> i haven't seen his new shop... i should drive by there and see it
[15:20:32] <seb_kuzminsky> Tom_itx: it's Mon Oct 17 - Sun Oct 23
[15:20:55] <jepler> I'll probably be there wed/thu through sunday night
[15:21:18] <Tom_itx> there's alot of road construction going on there now... just be aware
[15:21:26] <jepler> in wichita proper?
[15:21:31] <Tom_itx> yes
[15:21:40] <Tom_itx> they're fixing exits on that highway
[15:21:56] <Tom_itx> you can still get thru to west
[15:22:17] <Tom_itx> i'd almost recomend going to the next exit though since he moved to Harry St
[15:23:22] <jepler> I'll probably drive like an idiot to the old mpm1 location and then realize I don't know where the venue actually is
[15:23:25] <Tom_itx> taking SW BLVD South to Harry then back West
[15:23:38] <Tom_itx> it's not that far from his other place
[15:24:28] <Tom_itx> the 54/400 - 235 interchange is completely being redone getting rid of the old style cloverleaf
[15:24:40] <Tom_itx> which affects clear down to west st
[15:24:52] <Tom_itx> https://www.google.com/maps/@37.6722244,-97.3902417,15z
[15:25:22] <Tom_itx> get off at Southwest Blvd and go south to harry then back west a couple blocks
[15:25:32] <Tom_itx> is your best bet
[15:26:14] <Tom_itx> you can't get back on 54/400 from West right now
[15:26:20] <jepler> not relevant to anything we've been talking about but wow that is a tight bend in the railroad tracks by W 1st St N between Meridian and Seneca
[15:26:49] <Tom_itx> it's a local service track
[15:26:55] <Tom_itx> generally has parked trains on it now
[15:29:31] <Tom_itx> alot of those come from Vulcan (Occidental Chemical Corp) and park there until delivered or refilled
[15:29:48] <Tom_itx> ~55th & Ridge
[15:30:35] <Tom_itx> oh, if you're flying in.. we've got a new airport now too
[15:35:03] <Tom_itx> carry on...
[16:01:45] <jepler> ugh I can't believe they put the 3.3V / 5V switch on the edge where it's very bumpable
[16:02:02] <jepler> I wonder whether to fix it by a big blob of superglue on the switch itself, or by desoldering the switch and replacing it with a short jumper wire
[16:02:49] <jepler> http://ameridroid.com/products/xu4-shifter-shield
[16:02:51] <cradek> glue gun, not super glue (too thin)
[16:02:58] <cradek> but yeah, solder is better
[16:14:21] <KGB-linuxcnc> 05cradek/2.6/interp-startup 588b186 06linuxcnc 04. branch deleted * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=588b186
[16:16:04] <KGB-linuxcnc> 03John Morris 052.6 8b97e39 06linuxcnc 10src/emc/rs274ngc/rs274ngc_pre.cc interp: don't drop remap level at prog exit * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8b97e39
[16:16:04] <KGB-linuxcnc> 03John Morris 052.6 ca7c2de 06linuxcnc 10(6 files) Unit test for M30 and remapped command interaction * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ca7c2de
[16:28:45] <KGB-linuxcnc> 03John Morris 052.6 2e52efe 06linuxcnc 10src/emc/rs274ngc/interp_o_word.cc 03tests/interp/subs-follow-main/expected 03tests/interp/subs-follow-main/test.ngc 03tests/interp/subs-follow-main/test.sh Fix subs breaking when placed after main program * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2e52efe
[17:27:10] <linuxcnc-build> build #572 of 4017.5.deb-wheezy-armhf is complete: Failure [4failed shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4017.5.deb-wheezy-armhf/builds/572 blamelist: John Morris <john@zultron.com>
[18:18:07] <jepler> g++ 5.4 (ubuntu 16.04) on ARM seems to do much better on memory usage. I can build "-j8" on the xu4, with 2GB real ram + 1GB swap. even the big boost-ridden source files seemed to top out at 300MB "RES"
[18:19:56] <jepler> software opengl with "llvmpipe" makes the linuxcnc logo rotate nice and fast on the local display. still not sure how it would fare with a big glue-gun part program
[18:20:10] <jepler> (hm glue-gnu would be a great name for a software product in that market)
[18:21:17] <seb_kuzminsky> isnt glue *made* from things like gnus?
[18:21:25] <seb_kuzminsky> isn;t that a bit insensitive jepler
[18:21:31] <KGB-linuxcnc> 03Jeff Epler 05master 0763260 06linuxcnc 10scripts/travis-install-build-deps.sh travis: mk-build-deps can install the deps too * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=0763260
[18:27:10] <jepler> yeah but gnu-gun would also be a divisive software name
[18:33:18] <jepler> what other weird behaviors have we heard of on ubuntu 16.04? there's the boost double-free, which we worked around; and the axis y problem, which we worked around...
[18:35:17] <jepler> huh can't select axis letters by typing x, y, z...
[18:35:25] <jepler> `12 work
[18:36:32] <jepler> yuck works after alt-tabbing away and back
[19:00:09] <linuxcnc-build> build #3168 of 4006.deb-lucid-rtai-i386 is complete: Failure [4failed shell_2] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4006.deb-lucid-rtai-i386/builds/3168 blamelist: John Morris <john@zultron.com>
[19:00:40] <seb_kuzminsky> speaking of ubuntu problems
[19:00:40] <seb_kuzminsky> W: Failed to fetch http://old-releases.ubuntu.com/ubuntu/dists/lucid/Release.gpg Could not connect to old-releases.ubuntu.com:80 ( - connect (110: Connection timed out)
[19:11:42] <linuxcnc-build> build #573 of 4017.5.deb-wheezy-armhf is complete: Failure [4failed shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4017.5.deb-wheezy-armhf/builds/573 blamelist: John Morris <john@zultron.com>
[19:21:51] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15jepler opened pull request #158: packaging: stanza for ubuntu 16.04 (06master...06xenial) 02https://github.com/LinuxCNC/linuxcnc/pull/158
[19:46:33] <KGB-linuxcnc> 03Jeff Epler 05master 49a6ada 06linuxcnc Merge remote-tracking branch 'gh-jepler/xenial' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=49a6ada
[19:51:39] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15cradek closed pull request #158: packaging: stanza for ubuntu 16.04 (06master...06xenial) 02https://github.com/LinuxCNC/linuxcnc/pull/158
[20:36:21] <cradek> I guess github didn't notice my pushes matched those PRs
[20:36:59] <cradek> I used rebase, and also reordered one of them... can only ask for so much magical integration I guess
[20:47:58] <skunkworks> zlog
[21:04:59] <jepler> cradek: yes it looks at whether a ref is merged to master yet
[21:05:07] <jepler> cradek: so in that case you get to manually close them, as you seem to have done
[21:08:04] <jepler> er have you done?
[21:10:25] <cradek> no I didn't - I expected them to autoclose
[21:10:36] <cradek> also I didn't do the mergeup - I guess maybe that's required?
[21:10:51] <cradek> I put them both on 2.6 as he had PR'd