#linuxcnc-devel | Logs for 2014-12-15

[01:35:39] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.7 dac3a54 06linuxcnc 10docs/src/getting-started/Updating_LinuxCNC.txt docs: remove old 2.4->2.5 and 2.5->2.6 upgrade instructions * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=dac3a54
[01:35:40] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.7 8ce76bc 06linuxcnc 10docs/src/getting-started/Updating_LinuxCNC.txt 03docs/src/getting-started/images/upgrading-to-2.7.png docs: add 2.6->2.7 upgrade instructions * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8ce76bc
[02:04:10] <seb_kuzminsky> god i hate writing documentation
[02:36:54] <memfrob> are the foreign docs translated by someone or by google translate / similar?
[05:52:20] <jthornton> memfrob, the french and some of the spanish are translated by a person
[10:16:05] <mozmck> wow, looks like the build server's net connection is flaky
[10:17:15] <mozmck> I noticed this post: http://blog.machinekit.io/2014/11/jog-while-paused.html I forget what was not liked about Michael's solution?
[12:31:56] <seb_kuzm1nsky> oops
[12:43:30] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/v2_hallib 239f069 06linuxcnc 10(37 files in 10 dirs) hallib: support for system-wide halfiles * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=239f069
[13:16:35] <memfrob> The very first 3.16 ipipe patch is now available on my github page if anyone wants to try it out
[14:22:20] <seb_kuzminsky> sorry everyone for the join/leave noise... i had network trouble, it's been resolved, everything should be back to normal now
[14:23:17] <skunkworks> awesome - again - thanks for all your hard work
[14:23:33] <seb_kuzminsky> memfrob, JT-Shop, jthornton: i'm hoping to tackle the mess that is our documentation translation in the 2.8 timeframe, i've started doodling a fix in the branch called seb/master/po4a, if you want to take a look
[14:23:44] <seb_kuzminsky> skunkworks: :-)
[14:29:12] <memfrob> skunkworks: me or seb_kuzminsky?
[14:29:20] <skunkworks> Both!
[14:29:27] <memfrob> haha thank you!
[14:29:52] <memfrob> it is nice to finally be ahead of the xenomai team
[14:30:00] <skunkworks> well dad managed to get another cnc with a working control.. Damn it!
[14:30:29] <skunkworks> memfrob, I really want to test - but time seems to be sort at the moment..
[14:32:06] <seb_kuzminsky> memfrob: that's really cool
[14:32:09] <mozmck> hey seb_kuzminsky, we noticed the network trouble :) thought maybe you were getting buried in snow or something.
[14:32:15] <seb_kuzminsky> ugh
[14:32:16] <seb_kuzminsky> sorry
[14:32:20] <seb_kuzminsky> should be over now
[14:32:32] <mozmck> had problems myself yesterday.
[14:37:36] <mozmck> memfrob: is xenomai going to even keep updating IPIPE? with xenomai 3 they are just using preempt-rt
[14:38:23] <mozmck> which doesn't make any sense to me. I would think people would not need xenomai at all if they are using preempt-rt
[14:40:22] <seb_kuzminsky> the future of preempt-rt is looking uncertain too :-(
[14:40:24] <seb_kuzminsky> https://lwn.net/Articles/604695/
[14:40:57] <mozmck> Yeah, I saw that.
[14:41:34] <mozmck> Maybe that will inspire people to keep updating IPIPE
[15:11:55] <kwallace> skunkworks, are you going to be able to keep your new machines busy? Baby pictures would be nice too.
[15:22:42] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/v2_hallib 5636ed0 06linuxcnc 10(37 files in 10 dirs) hallib: support for system-wide halfiles * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=5636ed0
[15:27:37] <mozmck> I noticed this post: http://blog.machinekit.io/2014/11/jog-while-paused.html I forget what was not liked about Michael's solution? What is different with dgarr's moveoff?
[15:29:03] <Tom_itx> dgarr has several examples
[15:30:00] <mozmck> I've seen the videos, and it looks interesting, but I'm just curious what the basic difference is. Someone mentioned that the solution in machinekit was deficient, but I can't remember how.
[15:30:28] <Tom_itx> i haven't seen it work
[15:32:36] <Tom_itx> the video link you posted seems to do what would be expected
[15:33:18] <jepler> mozmck: I don't know anything about the implementation that was ultimately made in machinekit, because I don't follow that project from day to day.
[15:33:32] <jepler> mozmck: of course, they also aren't offering patches for inclusion in linuxcnc
[15:33:43] <jepler> mozmck: so it's silly to ask why we didn't take a hypothetical patch which they did not offer
[15:34:06] <Tom_itx> a little friction there?
[15:34:07] <jepler> there was some work done in 2013 before the split. this work can be found in our branch keyboard-jog-while-pause-joints-axes4
[15:34:37] <mozmck> hmm, I wasn't asking that, I was just curious to know the difference, because I thought someone had mentioned the machinekit solution having some problems.
[15:34:57] <jepler> compared to that work, dgarr's work is preferable because it doesn't affect any user who doesn't enable it. the keyboard-jog-while-pause-joints-axes4 branch changes the realtime motion controller which every user users
[15:35:00] <cradek> that video is a year old. contrary to my memory, it shows jogging with the regular AXIS controls. I don't know how that was accomplished. I couldn't find in their git the sim config the blog post talks about.
[15:35:01] <jepler> every user uses
[15:35:58] <jepler> one antipattern we have in the LinuxCNC task and motion controllers is: too many queues
[15:36:11] <mozmck> according to that blog post I linked, the code was actually done for linuxcnc, I assume in the branch you mention.
[15:36:20] <jepler> we have a queue for commands from user interfaces. a queue for mdi commands. a queue of motion commands waiting to be sent to realtime. a queue of motion commands in realtime.
[15:36:35] <jepler> a lot of the tough bugs we solve are due to these queues
[15:36:36] <Tom_itx> i haven't had a chance to try it yet but i think i will prefer dgarr's solution
[15:36:50] <mozmck> the video shows ja3 in the title actually
[15:36:57] <Tom_itx> i saw his demos he posted
[15:37:00] <jepler> so I would hate to take a patch that adds yet another kind of different queue, which is the first thing that stands out for me in this branch
[15:37:04] <jepler> +#define DEFAULT_ALT_TC_QUEUE_SIZE 100 // size of secondary motion queue
[15:37:29] <cradek> the recent patch I've seen would NOT have worked like the video (jogging with the usual jog controls)
[15:37:39] <cradek> I don't know how the video shows that
[15:37:46] <mozmck> ok, so maybe that's what I'm remembering.
[15:38:31] <seb_kuzminsky> i've played with dgarr's stuff, it does the thing people say they want: let you move the machine while a running program is paused
[15:38:42] <seb_kuzminsky> it'll be in 2.7.0~pre3, i'm really excited for that
[15:39:09] <Tom_itx> will it allow a tool offset to be set and then resume?
[15:39:15] <Tom_itx> that would be awesome...
[15:39:17] <seb_kuzminsky> nope
[15:39:35] <Tom_itx> as i see it, that's one of the main reasons for a pause in mid cut
[15:39:44] <cradek> me too, Tom_itx
[15:39:46] <seb_kuzminsky> i agree
[15:39:55] <Tom_itx> i know it's easy to talk about...
[15:40:16] <cradek> that will surely be part of the end goal, which neither of these stopgaps can handle
[15:41:58] <Tom_itx> i have never tried run from line either
[15:41:59] <skunkworks> tool lenght could be accomplished with it I think...
[15:42:08] <seb_kuzminsky> skunkworks: not as currently written
[15:42:27] <seb_kuzminsky> you have to clear all moveoff offsets (ie, jog back to where you started) before you un-pause
[15:42:30] <Tom_itx> diameter offset would be a big plus since the new tool isn't gonna measure the same
[15:42:54] <Tom_itx> or rather the ability to re'read the tool table parameters
[15:43:02] <skunkworks> seb_kuzminsky: ah.. I thought you could start it with an offset in place - it would just scold you.
[15:43:04] <cradek> can't make an omelet without opening all these cans of worms
[15:43:32] <Tom_itx> do you make custom omlets?
[15:43:33] <skunkworks> ewww
[15:43:58] <jepler> imagine an infinite barrel of monkeys, each opening cans of worms
[15:44:04] <cradek> haha
[15:44:08] <cradek> it'd look like irc
[15:44:09] <mozmck> thanks for the answers, not really important but I was curious. I'm happy for what is there.
[15:44:21] <seb_kuzminsky> skunkworks: hmm... now you've made me doubt myself
[15:44:22] <Tom_itx> i think most of us are
[15:45:06] <skunkworks> the original video that he posted I think allowed it - it may be how you have it setup..
[15:48:51] <mozmck> On another subject: does S code cause a pause in motion when it executes, and can I remap the S code and it not pause motion when it executes?
[15:52:11] <seb_kuzminsky> mozmck: i dont think S is a queue-buster, but it might make spindle-at-speed go false which would pause before the next feed move to let the spindle adjust
[15:53:01] <Tom_itx> what if you're doing sync'd spindle motion?
[15:53:05] <mozmck> hmm, is that true if I remap it as well? I need a code to send values from Gcode to a custom component.
[15:53:15] <Tom_itx> i suppose the S override is enabled in that case...
[15:53:43] <jepler> motion.spindle-at-speed IN BIT
[15:53:44] <jepler> Motion will pause until this pin is TRUE, under the following
[15:53:46] <jepler> conditions: before the first feed move after each spindle start
[15:53:49] <jepler> or speed change; before the start of every chain of spindle-syn‐
[15:53:52] <jepler> chronized moves; and if in CSS mode, at every rapid->feed tran‐
[15:53:55] <jepler> sition.
[15:54:02] <mozmck> Not using it for spindle control at all in this case, and I could use a different code if I have to. S is what we already use though.
[15:55:12] <jepler> I don't know enough about remap to be able to say
[15:55:43] <mozmck> ok, thanks. I will make a setup to do some testing.
[15:55:48] <mozmck> bbl
[15:57:25] <seb_kuzminsky> mozmck: if you dont care about spindle-at-speed, just set it True and forget about it
[15:57:34] <mozmck> so queue-buster means it will pause motion?
[15:57:43] <mozmck> seb_kuzminsky: I see, thanks!
[15:57:53] <seb_kuzminsky> it's... complicated, but sort of, yes
[15:58:16] <mozmck> where is the documentation for the main HAL components such as motion?
[15:58:41] <cradek> motion has a manpage
[15:58:54] <mozmck> ah, I see!
[15:58:58] <cradek> as do all (??) the hal components
[15:59:12] <Tom_itx> s/do/should
[15:59:26] <mozmck> even on the web under docs!
[15:59:34] <mozmck> ok, have to run, thanks for the help.
[18:49:36] <memfrob> mozmck: xenomai 2.6 hasn't been dropped yet
[18:50:52] <memfrob> seb_kuzminsky: thank you :) i took at look at porting IPIPE to 3.18 and it would take a few days but its mostly careful copypasta, not any re-write really
[18:51:20] <memfrob> the hardest part about 3.16 was the new assembly code stuff, thats why i needed one of the xenomai devs to fix the entry_32 and entry_64.S files
[20:45:12] <cradek> seb_kuzminsky: reading way back: sure I can
[21:49:49] <seb_kuzminsky> cradek: thanks! i think it'll be simpler if the install instructions can say just "install off this cd" instead of "install off this cd, then do the upgrade procedure"
[22:21:11] <seb_kuzminsky> just to annoy jepler:
[22:21:14] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a ab4fc27 06linuxcnc 10docs/src/Submakefile build with po4a v0.42, from wheezy * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ab4fc27
[22:58:09] <cradek> seb_kuzminsky: done, untested, path is binary27.....(.zsync)
[22:58:32] <cradek> I had good luck seeding with the other one
[23:02:15] <seb_kuzminsky> great! i'll test it out
[23:05:08] <seb_kuzminsky> i got bad status code 403
[23:05:29] <seb_kuzminsky> partway through the download
[23:07:11] <seb_kuzminsky> ah
[23:07:17] <seb_kuzminsky> -rw------- 1 emcboard pg167418 1187692544 Dec 16 2014 binary27.hybrid.iso
[23:07:24] <seb_kuzminsky> fixed
[23:07:48] <seb_kuzminsky> i_should_buy_a_faster_disk.jpg
[23:07:58] <cradek> oops
[23:08:35] <seb_kuzminsky> is it the same as the 2.6 iso except with 2.7.0~pre2?
[23:08:53] <seb_kuzminsky> ok now it's downloading
[23:09:01] <seb_kuzminsky> zsync is cool
[23:09:10] <cradek> yes and other package updates
[23:09:16] <seb_kuzminsky> right
[23:09:18] <seb_kuzminsky> perfect, thanks
[23:09:27] <cradek> welcome
[23:09:35] <cradek> thanks for making packages!
[23:09:42] <seb_kuzminsky> :-)
[23:10:01] <seb_kuzminsky> used 978345984 local, fetched 210429668
[23:10:52] <cradek> I'd've tested it, but I can't run qemu on my selectric terminal
[23:11:27] <seb_kuzminsky> heh
[23:12:38] <seb_kuzminsky> it boots to the live system
[23:12:47] <cradek> yay
[23:13:13] <seb_kuzminsky> and my VM has 20,000,000 ns jitter, but i'm ok with that
[23:13:19] <skunkworks> 2.7 livecd?
[23:13:22] <seb_kuzminsky> ye
[23:13:44] <seb_kuzminsky> err, way ahead of 2.7.0 i should note....
[23:14:30] <seb_kuzminsky> and the touchy/glavecdp sim config runs
[23:14:38] <seb_kuzminsky> this is great
[23:15:24] <cradek> super
[23:16:01] * seb_kuzminsky recklessly overwrites his wheezy machine with the new installer
[23:16:24] <cradek> eh, it'll be fine
[23:16:30] * seb_kuzminsky sometimes refers to himself in the third person
[23:16:34] <cradek> I should try it on my lathe
[23:16:50] <cradek> although I've been milling and not latheing lately
[23:17:23] <seb_kuzminsky> quick, somebody make a joke about booting the image off punch cards or something
[23:18:24] <cradek> I've never booted from punch cards (that can't be a thing) but I've certainly booted from tape
[23:19:18] <cradek> I have a dim memory of early hpux versions swapping to tape to install on low-ram machines
[23:19:30] <cradek> I'm only about half sure that wasn't a bad dream
[23:23:05] <seb_kuzminsky> it sounds terrible
[23:23:48] <seb_kuzminsky> i remember copying audio tapes containing pirated video games, and having like 80% of the copy work
[23:24:48] <cradek> heh, copied on regular audio equipment?
[23:25:57] <seb_kuzminsky> yeah, my parents had a stereo with not one but two casette decks
[23:26:02] <seb_kuzminsky> it was deluxe
[23:26:07] <cradek> I still have one!
[23:26:17] <cradek> I sometimes use it to make tapes for ... my car
[23:26:24] <seb_kuzminsky> do all your tapes sound all warbly
[23:26:26] <seb_kuzminsky> ?
[23:26:30] <cradek> no, they're fine
[23:26:38] <cradek> maybe that's why your games didn't run, haha
[23:26:45] <cradek> needed new rubber bands
[23:26:59] <seb_kuzminsky> they should invent a bluetooth audio receiver built into a casette, powered by the tape drive motor
[23:27:04] <seb_kuzminsky> for, you know, cradek
[23:27:21] <cradek> what use would I have for that? I've got all these tapes.
[23:27:43] <cradek> sometimes my life borders on ridiculous
[23:27:58] <cradek> fortunately I like it that way just fine
[23:28:03] <seb_kuzminsky> you could send yourself a voicemail of you playing your audio tapes, then listen to them in the car
[23:29:03] <seb_kuzminsky> the install worked
[23:29:08] <cradek> cool
[23:29:17] <seb_kuzminsky> not that i'm surprised
[23:29:29] <cradek> if you're bold you might make sure I have the (commented) buildbot sources spelled right
[23:29:39] <seb_kuzminsky> oh yeah, good idea
[23:29:55] <cradek> I checked them, but you know how it is
[23:30:16] <seb_kuzminsky> the 2.7 part of the buildbot's going to change one of these weeks, that's why i havent announced it yet
[23:30:35] <seb_kuzminsky> in 2.6 and older we have linuxcnc.deb (for rtai) and linuxcnc-sim.deb (for vanilla)
[23:31:19] <seb_kuzminsky> err, i mean the deb components are 2.6-rt (for rtai) and 2.6-sim (for vanilla)
[23:31:33] <seb_kuzminsky> in 2.7 i want 2.7-rtai for rtai and 2.7-uspace for rt-preempt and vanilla
[23:31:52] <seb_kuzminsky> but i need to futz with the buildbot to make the old and the new ways work right somehow
[23:32:06] <seb_kuzminsky> i like that you put git on the install iso
[23:32:42] <cradek> once you give up on making it fit on a cd, it gets easy
[23:33:05] <seb_kuzminsky> in that case can you add screen to the next one? :-)
[23:33:07] <cradek> and I've heard very little complaining about that decision
[23:33:20] <seb_kuzminsky> i've heard none
[23:33:30] <seb_kuzminsky> i think the age of cds may finally be behind us
[23:34:08] <cradek> sure, screen added
[23:34:57] <seb_kuzminsky> yay
[23:35:48] <seb_kuzminsky> when i ran "apt-get install screen" it ran a bunch of triggers ("Processing triggers for man-db ...", etc), then it said
[23:35:59] <seb_kuzminsky> insserv: warning: script 'realtime' missing LSB tags and overrides
[23:36:06] <cradek> hmm
[23:36:11] <seb_kuzminsky> i don't think i've seen that before, but maybe i've just missed it
[23:36:34] <cradek> are they screwing with init yet? surely not in the stable version
[23:36:37] <seb_kuzminsky> the realtime script works, because linuxcnc starts just fine
[23:37:58] <seb_kuzminsky> /sbin/init is from sysvinit.deb
[23:41:55] <seb_kuzminsky> it's got mesaflash 3.1, cool
[23:42:17] <seb_kuzminsky> and truetype-tracer 4.1
[23:42:27] <seb_kuzminsky> the buildbot apt source lines are working
[23:42:56] <seb_kuzminsky> this looks great cradek
[23:43:21] <seb_kuzminsky> it's kind of magical when everything works right
[23:43:54] <cradek> I sure like when the work is already done, and I can just tweak it and rebuild
[23:44:21] <cradek> if you want to rename the file and fix up links to it, please do
[23:44:29] <cradek> the filenames are silly
[23:44:41] <seb_kuzminsky> they are, a little
[23:44:58] <seb_kuzminsky> there are no links yet, right?
[23:45:01] <seb_kuzminsky> to the iso i mean
[23:45:09] <cradek> not to the 27
[23:46:05] <seb_kuzminsky> i won't rename the 2.6 install iso but i will rename the 2.7 one
[23:46:14] <cradek> perfect
[23:46:19] <seb_kuzminsky> linuxcnc-2.7-wheezy.iso
[23:46:32] <cradek> ok
[23:49:00] <cradek> I think you have to edit the beginning of the .zsync file if you rename it
[23:49:03] <cradek> or something
[23:49:41] <seb_kuzminsky> thanks for the heads up
[23:49:56] <seb_kuzminsky> whoops, note to self, dont try to edit the iso by mistake
[23:51:42] <seb_kuzminsky> is it a bad sign if i accidentally run "vi mumble.iso" on snotra, and it hangs for a minute and then kills my ssh connection?
[23:52:49] <seb_kuzminsky> editing the .zsync file works much better
[23:57:24] <seb_kuzminsky> hmm, well i renamed the iso and zsync file, but now it's ignoring my local copy and re-downloading the whole thing
[23:57:33] <seb_kuzminsky> (i renamed my local copy to match)
[23:57:48] <cradek> use -i the first time?
[23:58:06] <cradek> I don't know how it keeps the two things connected by default
[23:58:20] <cradek> you'd think it would be "take off the .zsync extension" but it seems to be more complicated than that