#linuxcnc-devel | Logs for 2016-08-16

[00:00:15] <mozmck> And it ran fine for 3 passes here. Late here - off to bed.
[01:56:06] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15zultron commented on issue #138: @SebKuzminsky, welcome back!... 02https://github.com/LinuxCNC/linuxcnc/issues/138#issuecomment-240014561
[06:40:42] <skunkworks> zlog
[09:39:36] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15jmkasunich opened issue #143: mux_generic man page quibbles 02https://github.com/LinuxCNC/linuxcnc/issues/143
[10:18:35] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15cradek opened pull request #144: Epp parport g540 (06master...06epp-parport-g540) 02https://github.com/LinuxCNC/linuxcnc/pull/144
[10:19:35] * cradek is inspired by jmkasunich to fiddle with github
[10:32:28] <KGB-linuxcnc> 03Chris Radek 05master d1fe691 06linuxcnc Merge branch 'execute_finish_in_py_remap-lcnc' of https://github.com/zultron/machinekit into zultron-execute_finish_in_py_remap-lcnc * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=d1fe691
[10:34:04] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15cradek closed pull request #135: Execute finish in py remap lcnc (06master...06execute_finish_in_py_remap-lcnc) 02https://github.com/LinuxCNC/linuxcnc/pull/135
[10:34:15] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15jepler commented on issue #142: @jmkasunich because of the way github pull requests work, the second commit you added to your master branch, "halstreamer: allow comments in input stream" becomes a part of this pull request. To have multiple outstanding pull requests with different contents, you need to use a different branch name for each one. Each branch should start at linuxcnc "uptream" master.... 02https://github.
[10:36:35] <cradek> I must've missed a spot where I could name the pull request
[10:39:38] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15cradek commented on issue #135: @zultron Thank you for your continuing work on remap. 02https://github.com/LinuxCNC/linuxcnc/pull/135#issuecomment-240133165
[10:54:49] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15SebKuzminsky commented on issue #138: Yeah, that "Abort implies M2" behavior is way too heavy handed, and is the cause of some of the bugs that cropped up as a result of that work.... 02https://github.com/LinuxCNC/linuxcnc/issues/138#issuecomment-240138351
[10:58:16] <skunkworks> Hmm - it almost seems like the single index interpolated cut better than the 100 line not interpolated. (just looking at the shavings)
[10:58:27] <skunkworks> but maybe that is just me.
[10:58:50] <skunkworks> *shavings coming off the tool
[10:59:36] <seb_kuzminsky> that's strange
[11:00:08] <seb_kuzminsky> i wonder if there's high-frequency wobble in the spindle speed, that gets smoothed out by reducing the sample speed from 100 Hz to 1?
[11:00:17] <skunkworks> the finished product looked pretty identical
[11:00:45] <seb_kuzminsky> it'd be interesting to see a halscope trace of interpolated spindle speed from the 1 Hz signal and uninterpolated from the 100 Hz signal
[11:00:56] <skunkworks> seb_kuzminsky, seems like more inconsistant shavings from path to path compared to the single index interpolated.
[11:01:37] <seb_kuzminsky> huh, it changes from pass to pass, but it stays consistent within a single pass? so much for my theory
[11:01:46] <skunkworks> right
[11:02:04] <skunkworks> again - just going from watching the video... So who knows.
[11:03:36] <skunkworks> aluminum was from the same shaft - so no difference there
[11:03:49] <seb_kuzminsky> the retract is really slow in the 100 line video
[11:03:59] <seb_kuzminsky> i mean the return part of the pass
[11:04:10] <seb_kuzminsky> shouldn't that happen at rapid speed?
[11:04:10] <skunkworks> really? same program
[11:04:51] <seb_kuzminsky> i'm not saying 'in comparison to the index-only video', i just mean it's slow looking at it all by itself
[11:05:11] <skunkworks> oh
[11:05:15] <cradek> I think it's threading at almost-rapid speed
[11:05:15] <seb_kuzminsky> the index-only one is slow too
[11:05:20] <seb_kuzminsky> heh ok
[11:06:28] <seb_kuzminsky> i'm used to cutting at like 1/10 the rapid speed and having the machine act all spastic when the tool's out of the work
[11:06:35] <skunkworks> yes - it is threading at around 27ipm and the max is 30 currently
[11:07:30] <seb_kuzminsky> one of these winters i really need to tune up my little 7x14 and mount the servos on it
[11:08:35] <cradek> I guess I don't have a youtube video of the hnc threading
[11:09:20] <skunkworks> I did a 9tpi I think - but that is about the limit for that machine
[11:09:20] <cradek> did you see the photos of ed nisley's new "lathe"?
[11:10:39] <seb_kuzminsky> i've been aware he got a little 7xmumble like mine and has been doing his usual thorough inspection and tune-up on in
[11:11:37] <seb_kuzminsky> i put a link to his blog posts in the 'notes' file for my lathe cnc project
[11:14:09] <cradek> https://softsolder.files.wordpress.com/2016/06/dsc08808-lms-mini-lathe-carriage-flat-way-detail.jpg
[11:14:53] <cradek> the part that isn't the ways looks surface-ground
[11:15:11] <cradek> it's not clear what happened to the part that is
[11:15:34] <cradek> (I'm glad I got a sherline instead of one of these, way back when)
[11:17:38] <KGB-wlo> push to master branch: http://linuxcnc.org/
[11:17:41] -linuxcnc-github:#linuxcnc-devel- [13wlo] 15jepler pushed 1 new commit to 06master: 02https://github.com/LinuxCNC/wlo/commit/f5c084d924461aef2186be3023e0c91b2ee700d6
[11:17:41] -linuxcnc-github:#linuxcnc-devel- 13wlo/06master 14f5c084d 15Jeff Epler: new showcase...
[11:18:59] <seb_kuzminsky> thanks jepler
[11:19:02] <cradek> yay!
[11:37:44] <KGB-linuxcnc> 03John Thornton 052.7 1489c87 06linuxcnc 10docs/man/man9/mux_generic.9 Docs: fix issue #143 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=1489c87
[11:39:07] * skunkworks needs to start doing more of that...
[11:39:51] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15jethornton commented on issue #143: Fixed with commit 1489c8 on 2.7 branch 02https://github.com/LinuxCNC/linuxcnc/issues/143#issuecomment-240152614
[11:40:27] <seb_kuzminsky> jthornton: if you say "fixes #143" in the commit message, github will comment & close the issue for you
[11:40:41] <jthornton> ok thanks
[11:40:46] <jthornton> didn't know
[11:41:22] <jthornton> so fixes and the number is the magic word
[11:41:38] * jthornton goes back to work
[12:27:59] <skunkworks> wow - I should have proofread my email.
[14:04:44] <jepler> skunkworks: I find I get a lot less e-mail sent when I do that
[14:04:55] <jepler> why not just spend more time on IRC where noby cares about your typos
[14:09:12] <skunkworks> I do - I do
[14:22:30] <KGB-linuxcnc> 03Jeff Epler 05master 2211e1c 06linuxcnc Merge branch 'master' of https://github.com/jmkasunich/linuxcnc * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2211e1c
[14:27:37] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15cradek closed pull request #142: limit3.comp: add 'in-limit' output pin (06master...06master) 02https://github.com/LinuxCNC/linuxcnc/pull/142
[14:37:26] <skunkworks> hmm - my clock doesn't keep time when it goes to sleep
[14:37:32] <skunkworks> *laptop
[14:37:48] <skunkworks> oh - never mind. it takes a second
[16:18:56] <jepler> here's a thing I've been toying with: https://emergent.unpythonic.net/files/sandbox/git-merge-pr
[16:19:07] <jepler> it lets me just 'git merge-pr 144' e.g.
[16:19:25] <jepler> after a modest amount of setup. h/t seb for the lead on the "jq" program.
[16:21:46] <seb_kuzminsky> ermagerd, work just brough in chocolate chip espresso cookies to the break room
[16:22:05] <jepler> seb_kuzminsky: I'll be right there, save one for me
[16:24:37] <jepler> > Intel Corp., the world’s biggest semiconductor maker, said it’s licensing technology from rival ARM Holdings Plc, a move to win more customers for its business that manufactures chips for other companies.
[16:24:44] <seb_kuzminsky> that merge-pr script looks handy for people like us, who are only dipping one toe in the github water
[16:25:49] <jepler> huh no idea what ARM's licenses are like, but I'm surprised Intel needs to own one to manufacture a design from Company X that will then be sold by Company X.
[16:26:37] <jepler> but hey I guess if ARM's license with X said that X could only send their work to a licensed fab, that has now paid off for ARM with steamships full of money from Intel.
[16:32:02] <seb_kuzminsky> jepler: when you "git pull $cloneurl $ref" like that, does it give you a nice merge message like "Merge remote-tracking branch 'pkmcnc/master'"? it can't, right?
[16:32:34] <jepler> $ git log --oneline -1
[16:32:34] <jepler> b127db0 Merge branch 'epp-parport-g540' of https://github.com/cradek/linuxcnc
[16:33:05] <seb_kuzminsky> noice
[16:33:10] <jepler> that's git's magic, not mine
[16:34:55] <jepler> tbh I don't know why github recommends the more complicated sequence of operations when you "view manual merge instructions" or whatever the faux-link is called
[16:38:11] <seb_kuzminsky> rtai.org published a new kernel patch for 4.1.18 that includes some locking changes (compared to the 3.18 patch i was testing a month or two ago)
[16:38:21] <seb_kuzminsky> i'm gonna build it and take it for a spin
[17:19:22] <jepler> debian seems to have missed the boat again, 4.9 is announced as an LTS kernel but Debian announced months ago they planned to pick 4.10 for stretch
[17:19:26] <jepler> sigh, oops, etc
[17:21:26] <seb_kuzminsky> :-(
[17:21:46] <seb_kuzminsky> that announcement seemed really premature
[17:22:58] <jepler> I understand where they were coming from: by waiting just a few days they could have a newer kernel
[17:23:24] <jepler> but having an lts kernel has to be better than having a month-newer kernel, or whatever the kernel release cycle actually is
[17:23:34] <seb_kuzminsky> that they have to support themselves until everyone abandons it for the bpo version
[17:24:25] <jepler> hopefully stretch works properly with a 4.9 kernel, much more likely to see RT patches of whatever stripe for LTS kernels
[17:25:27] <seb_kuzminsky> or whatever the lts kernel after 4.10 is
[17:26:26] <jepler> seb_kuzminsky: I'd like to talk sometime about buildbot and uspace
[17:26:50] <jepler> now / tonight aren't great for me, maybe there's some evening or weekend day in the next week that we could pick and plan to both be on IRC?
[17:27:01] <seb_kuzminsky> yeah
[17:27:15] <seb_kuzminsky> i owe you several buildbot fixes & improvements
[17:27:21] <jepler> I dunno about "owe"
[17:28:13] <jepler> but I think that with only your work expended *cough* buildbot could be making packages that include uspace+rtai and maybe the number of uspace builders could be cut in half, since uspace and sim are the same thing
[17:28:46] <seb_kuzminsky> tonight & tomorrow are bad for me, friday's bad, so how about thursday eve?
[17:29:05] <jepler> sure why not
[17:29:15] <seb_kuzminsky> around 9 mountain?
[17:29:49] <jepler> earlier would be better for me, but I can do 9
[17:30:02] <seb_kuzminsky> 8:30?
[17:30:05] <jepler> sure
[17:30:10] <seb_kuzminsky> ok
[17:30:34] <jepler> "see" you then
[17:30:52] <seb_kuzminsky> alrighty
[17:31:05] <Guest29013> I confirmed on second Core 2 Duo machine that Preempt_RT/hm2-eth work well with 4.1 kernel but poorly with 4.4 or >
[17:31:06] <Guest29013> (no problem with 4.4 on newer machines though)
[17:31:59] <seb_kuzminsky> 15:12:13 -!- PCW is now known as Guest29013
[17:32:53] <Guest29013> these do work well as hm2_eth hosts:
[17:32:55] <Guest29013> http://www.ebay.com/itm/HP-COMPAQ-dc7900-ULTRA-SLIM-DESKTOP-CORE-2-DUO-3-00GHZ-2GB-RAM-USFF-W-ADAPTER-/311669115029?hash=item4890ed4895:g:6mMAAOSwhOdXog7Y
[17:34:52] <jepler> there are a pile of core2 full-size tower desktops in cradek's office, free if you can get him to carry it out to your car. don't recommend pulling that trick unaccompanied, the receptionist is on to you.
[17:35:56] <pcw_mesa> ran over the weekend with Peppa Pig compilations as torture...
[17:36:51] <pcw_mesa> Still have no idea why 4.4 has such bad latency on those
[17:37:05] <jepler> not familiar with that, but a quick screen-cap view shows me I would be tortured by it
[18:00:56] <pcw_mesa> Yes Peppa Pig torture is broad-band
[18:31:31] <Tom_itx> any particular eide parport card that works with lcnc?
[18:31:52] <Tom_itx> err pcie
[18:33:38] <seb_kuzminsky> i think they all work for bitbanging, and only some work for epp
[18:33:46] <seb_kuzminsky> pcw_mesa might know of some epp ones
[18:33:50] <seb_kuzminsky> bbl
[18:33:56] <Tom_itx> thanx
[18:36:59] <Tom_itx> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?EMC2_Supported_Hardware
[18:37:09] <Tom_itx> found that but don't know how recent / accurate it is
[18:47:05] <pcw_mesa> I haven't checked any PCIE parallel cards (or looked at parallel anything for a while)
[18:59:07] <Tom_itx> can you plug a pcie in a pci slot?
[19:00:25] <jepler> Tom_itx: no
[19:00:26] <Tom_itx> doesn't look like it...
[19:48:35] <skunkworks> zlog
[21:30:44] <jepler> whee jon e got his probe to do something!
[21:52:19] <jepler> seb_kuzminsky: looks like the buildbot is stuck. 1402.rip-wheezy-rtpreempt-i386
[21:52:22] <jepler> (offline, plus 1
[21:52:24] <jepler> )
[22:53:32] <seb_kuzminsky> thanks jepler, i poked it
[22:53:48] <seb_kuzminsky> in that special spot
[22:53:54] <seb_kuzminsky> now things are going again