Back
[09:31:41] <KGB-linuxcnc> 03Jeff Epler 052.7 8e7caf5 06linuxcnc 10debian/configure 10src/hal/hal_priv.h Merge remote-tracking branch 'origin/2.6' into 2.7 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8e7caf5
[09:31:41] <KGB-linuxcnc> 03Jeff Epler 052.7 3576acb 06linuxcnc Merge remote-tracking branch 'origin/jepler/arm-atomic' into 2.7 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=3576acb
[09:31:41] <KGB-linuxcnc> 03Jeff Epler 05master 8b989a1 06linuxcnc 10docs/src/gui/gmoccapy.txt 10scripts/linuxcnc.in Merge branch '2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8b989a1
[09:32:32] <seb_kuzminsky> aww yea
[09:35:45] <jepler> if we have an embarassingly bad 2.7.0 release, this can be the retroactive theme when we release 2.7.1:
https://www.youtube.com/watch?v=fGl508qjiSM
[09:39:52] <seb_kuzminsky> and our users be like:
https://www.youtube.com/watch?v=rxgWHzMvXOY
[09:42:13] <jepler> d'awwww that's so sweet
[09:42:52] <seb_kuzminsky> that video gives me diabeetus
[09:46:25] <cradek> I encountered this on the local weird hippie radio station this morning:
https://www.youtube.com/watch?v=2GHyLhbdzN0
[09:50:21] <seb_kuzminsky> i have a repro of the intermittent dropped mdi problem in the t0 tests
[09:50:26] <seb_kuzminsky> it's in task :-/
[09:50:27] <seb_kuzminsky> again :-/
[09:50:36] <cradek> yargh
[09:50:42] <seb_kuzminsky> it's in 2.6 & 2.7
[09:50:50] <seb_kuzminsky> oops forgot the :-/
[09:50:55] <cradek> :-/
[10:18:11] <KGB-linuxcnc> 03John Thornton 052.7 e3f7773 06linuxcnc 10(10 files in 4 dirs) Docs: more work cleaning up the anchors * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=e3f7773
[10:18:11] <KGB-linuxcnc> 03John Thornton 052.7 def4178 06linuxcnc 10(17 files in 6 dirs) Docs: more work cleaning up anchors * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=def4178
[10:18:11] <KGB-linuxcnc> 03John Thornton 052.7 0b8a3a1 06linuxcnc 10(9 files in 3 dirs) Docs: more work on anchors and links * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=0b8a3a1
[10:19:10] <jthornton> off to a customer while buildbot chews on that
[10:28:14] <skunkworks> smop!
[12:46:11] <linuxcnc-build> build #2511 of 1301.rip-precise-rtai-i386 is complete: Failure [4failed git] Build details are at
http://buildbot.linuxcnc.org/buildbot/builders/1301.rip-precise-rtai-i386/builds/2511 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[12:46:12] <linuxcnc-build> build #3317 of 0000.checkin is complete: Failure [4failed] Build details are at
http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/3317 blamelist: Sebastian Kuzminsky <seb@highlab.com>
[12:47:52] <seb_kuzminsky> boo:
http://highlab.com/~seb/linuxcnc/precise-rtai-i386-panic.png
[13:00:08] <skunkworks> is that new?
[13:08:55] <seb_kuzminsky> it's happened before with the 3.4-9-rtai-686-pae kernel, always while running the test suite
[13:08:59] <seb_kuzminsky> it's rare
[15:04:26] <jepler> huh, jump to NULL
[15:04:29] <jepler> that's not going to work out
[15:04:46] <jepler> some days I think I should port uspace to rtai's lxrt
[15:04:51] <jepler> then we could be rid of kernel modules forever
[15:09:16] <skunksleep> Could you do software stepping with lxrt?
[15:09:33] <jepler> I don't know.
[15:09:43] <jepler> I don't know which builds we've made of rtai support lxrt
[15:19:53] <jepler> Chris Lesiak is on the ball
[15:21:27] <seb_kuzminsky> yeah he's sharp
[15:22:32] <seb_kuzminsky> our 3.4-9-rtai kernel has lxrt:
http://paste.ubuntu.com/12002444/
[15:23:17] <jepler> wow 3.4 is old
[15:23:23] <jepler> </whine>
[15:28:49] <seb_kuzminsky> in theory it should be easier to make new rtai kernels now
[15:29:05] <seb_kuzminsky> the sticking point is finding one that:
[15:29:09] <seb_kuzminsky> * is supported by rtai
[15:29:20] <seb_kuzminsky> * is supported by debian's kernel packaging
[15:29:26] <seb_kuzminsky> * works
[15:30:10] <seb_kuzminsky> but instead of working on that, or anything else, imma go have lunch
[15:30:36] <kwallace> Has anyone made a G47 like this for LinuxCNC? :
http://gcodes.net/haas-g47-text-engraving/
[15:32:22] <kwallace> One tricky bit might that the text in the command "G47 P0 (TEXT TO ENGRAVE)" will be taken as a comment.
[15:32:35] <cradek> that's pretty sweet, but no
[15:32:50] <cradek> I just use truetype-tracer to generate gcode
[15:33:44] <cradek> in general, I don't agree with putting everything into the control. we have a nice multitasking general purpose computer *right there*.
[15:34:21] <cradek> with only a bit of perversion you could even implement it as an AXIS input filter
[15:35:20] <kwallace> I got a request to do a serial number thing of some sort. Haas' G47 does serial numbers.
[15:37:12] <cradek> ugh, the "P values to engrave specific characters" awfulness is a good reason to do this outside of gcode
[15:38:24] <cradek> wish we could have a gcode command that runs a shell command that spits gcode to its stdout
[15:38:37] <cradek> that's the flexible unix way to do this and a zillion other things
[15:39:52] <cradek> oh hey
[15:40:04] <cradek> M100 can already run a program
[15:40:22] <cradek> then O-call can open a gcode file at a predetermined filename
[15:40:33] <cradek> don't those two building blocks give you what you want?
[15:46:41] <kwallace> I don't know what I want.
[15:51:57] <seb_kuzminsky> cradek: that should work for numeric serial numbers that can be passed as P-word arguments to the M-code, but not for arbitrary text strings
[15:58:52] <cradek> for serial numbers you can't hardcode them in the gcode anyway.
[15:59:49] <seb_kuzminsky> #123 = [#123 + 1]
[15:59:53] <seb_kuzminsky> m100 p#123
[16:00:21] <seb_kuzminsky> what range of gcode parameters is saved in our .var file?
[16:01:37] <seb_kuzminsky> our docs say "currently 5161-5390"
[16:01:54] <seb_kuzminsky> but the docs definitely sound like they're hedging their bets on this topic
[16:02:11] <cradek> any numbers you add to the varfile get saved from then on
[16:02:31] <cradek> so yeah, that's a not-terrible-at-all approach, if it works
[16:03:26] <seb_kuzminsky> no way! if you can tell it to save additional parameters by just adding them to the var file, that's great
[16:03:34] <andypugh> You can
[16:03:42] <cradek> good luck writing "SN: %08d" in gcode
[16:04:04] <cradek> although I'm sure that's how it works, I wonder if it's actually documented
[16:04:56] <seb_kuzminsky> i dont see it documented in my casual glance here:
http://linuxcnc.org/docs/2.7/html/gcode/overview.html#_parameters
[16:05:55] <seb_kuzminsky> huh, down in the "numbered parameters" section, the G92 offset parameters are not listed as persistent, but surely they are
[16:06:39] <seb_kuzminsky> oop, yep, the addition of user-selected params to the persistence list *is* documented
[16:06:48] <cradek> yay!
[16:06:51] <seb_kuzminsky> http://linuxcnc.org/docs/2.7/html/gcode/overview.html#_numbered_parameters_persistence
[16:06:54] <seb_kuzminsky> yay :-)
[16:07:04] <cradek> kinda too bad that was not extended for named parameters
[16:08:26] <cradek> > You can see from the following figure that the output is not what you might expect.
[16:08:29] <cradek> hahaha
[16:09:31] <cradek> but that is what I expected!
[16:10:16] <seb_kuzminsky> i expected you to say that
[16:10:52] <cradek> our polar mode is so much awesomer than what my terrible bridgeport had
[16:11:18] <seb_kuzminsky> i never use polar mode, maybe i should try it more
[16:11:19] <cradek> I bet very few people have ever used it, but it's so neat
[16:11:21] <cradek> dude!
[16:11:47] <seb_kuzminsky> i built big ugly equations with SIN[] and COS[] instead :-/
[16:12:37] <cradek> well that ability is pretty neat, too
[16:13:00] <andypugh> I use polar mode in MDI quite often for bolt circles
[16:13:02] <cradek> I think polar with repeating canned cycles is particularly nice
[16:13:25] <cradek> yay :-)
[16:14:28] <andypugh> I wonder if Touchy users use it more than most, as it is there looking mysterious in the GUI?
[16:14:50] <cradek> it does stick out a bit
[16:15:13] <cradek> actually I think we might be the two touchy users
[16:15:32] <cradek> which is fine with me. it won't grow any unwanted features that way.
[16:15:43] <cradek> (oops, too honest?)
[16:16:38] <andypugh> Incidentally I have noticed that most of the Vismach demos start off in birds-eye-view which isn’t very clear. It is also not clear how to rotate the view to a more suitable angle. Is it worth adding a viewpoint to those that lack it?
[16:17:08] <andypugh> I am a bit concerned with the way Gmoccapy is going.
[16:17:11] <cradek> does it have that?
[16:17:16] <cradek> if so, sure, why not
[16:17:24] <cradek> concerned how?
[16:17:32] <andypugh> One at a time. :-)
[16:17:55] <andypugh> Vismacj takes a lat and long number as optional paramaters to main()
[16:19:11] <cradek> thinking a little bigger: maybe defaulting to 1,1,-1 would be better than 0,0,1
[16:19:11] <andypugh> For example: main(model, tooltip, work, 150, lat=-75, lon=215)
[16:19:42] <cradek> or 1,1,1 or whatever makes sense
[16:19:43] <andypugh> That too
[16:20:45] <andypugh> As for Gmoccapy, it seems that whenever Norbert sees a feature he wants he adds it to Gmoccapy, rather than considering non-gmp users
[16:21:15] <cradek> yes I suspect it has things that don't belong in the gui
[16:21:25] <andypugh> So, he has added a tool changer, and auto tool-length touch-off and a bunch of other stuff that I am not sure belong in a specific gui
[16:22:02] <cradek> what do you mean added a tool changer?
[16:22:09] * seb_kuzminsky blinks
[16:22:14] <andypugh> I am unclear on the details.
[16:22:52] <andypugh> It may just be a built-in version of hal_manualtoolchange
[16:23:00] <seb_kuzminsky> one of the things i like about linuxcnc is that we dont all have to agree on what the gui should look like
[16:23:00] <seb_kuzminsky> bbl, meeting
[16:23:01] <cradek> sounds like it's not a question of appearance
[16:23:34] <cradek> I admit I haven't tried to run gmocappy for a year or more. I don't know anything about it.
[16:23:48] <cradek> dangit, and I always spell it wrong
[16:23:48] <mozmck> I do like that myself seb. I haven't tried to use Gmoccapy seriously, but I couldn't stand the interface the little I played with it.
[16:25:29] <andypugh> I “approve” of the way that Touchy needs physical buttons to be pressed to make things happen.
[16:25:36] <mozmck> you left out the 'r' ;)
[16:25:55] <cradek> it's always bothered me a bit how that's not true for homing
[16:26:30] <cradek> not enough to figure out a new gui paradigm to fix it, though
[16:27:45] <cradek> I like touchy a lot. I think it has some really good ideas that are in none of the other UIs. It makes me sad that people discount it because it's utilitarian (i.e. ugly).
[16:28:33] <andypugh> Hmm,
http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Gmoccapy#Auto_Tool_Measurement reading the details it seems to be largely a normal remap with the GUI twiddling some pins.
[16:28:53] <andypugh> Maybe I am misinterpreting what gets described as a Gmoccapy “feature”
[16:29:31] <skunkworks> cradek: maybe use some of moses fancy buttons!
[16:30:39] <cradek> my original design was to have no english text at all, all icons, I even collected/studied pictures of commercial controls to see what they used
[16:30:42] <cradek> it fell by the wayside
[16:30:56] <cradek> even the standard icons are really opaque
[16:31:20] <cradek> pictures of paper tape and a hand with its finger pointing and all sorts of crap like that
[16:31:57] <cradek> can be used by anybody, no i18n needed = pie in the sky
[16:34:09] <andypugh> The LinuxCNC “Features” interface has some purrty buttons.
[16:34:11] <mozmck> I haven't really looked at touchy - I probably should! I don't have a touchscreen and have not really targeted one.
[16:34:42] <mozmck> My lightbutton is pretty configurable, but the look could be improved I'm sure.
[16:34:44] <Roguish> here's a pretty nice touch screen for touchy:
http://electronics.woot.com/plus/benq-65-full-hd-2-point-touch-led-display-1?ref=cnt_wp_8
[16:34:46] <cradek> andypugh: I should have realized that wasn't my forte and tried to team up with someone
[16:35:33] <andypugh> Is “Features” mainlined yet?
https://www.youtube.com/watch?v=91Eq75exU9o
[16:35:46] <mozmck> Roguish: If they move the decimal point left 3 places I might be interested.
[16:36:03] <cradek> well fwiw, "it makes me sad that people discount it because of looks" is not the same as "I wish it looked different"
[16:36:41] <cradek> it can already use all the various gtk themes, which is a very good feature you'd quickly lose
[16:37:11] <cradek> that's part of making it work well on any screen size
[16:37:56] <cradek> tormach's gui that uses images for everything, and therefore breaks all screen sizes but one, is the counterexample and I find that to be a very bad design.
[16:38:24] <cradek> (fortunately I'm not in marketing or sales)
[16:39:07] <Roguish> cradek: touchy is nice 'cause it's pretty clean. not a lot of crap on the screen. JThorton has a version with a gremlin tab which is really nice.
[16:39:33] <jthornton> I like Touchy and use it on my BP mill, the only thing I don't like is to have to touch the jog button after doing anything else
[16:39:39] <andypugh> I think I will add a “Features” tab and see how tha works out
[16:39:41] <cradek> yes the ability to add custom tabs is nice
[16:41:01] <cradek> jthornton: I wonder if that could be improved somehow
[16:41:04] <mozmck> In my gscreen based gui, I made it go to jog mode anytime no widget requiring keyboard input is selected.
[16:41:06] <jthornton> one thing that is so nice about Touchy when you start it up you have just what you need estop, power on, home all
[16:41:11] <Roguish> gmoccapy is ok. some of those 'features' are handy like the tool change and touch off stuff. maybe they're just remaps, but for us that don't want to struggle with writing remaps is handy. but the screen is way too busy
[16:41:53] <jthornton> be nice to have a setting or pin to change that, I have a selector switch for axis, increment and a MPG so it is all hardware
[16:41:53] <mozmck> Roguish: I don't understand why you have to jump through so many hoops to load a gcode program.
[16:42:24] <andypugh> Hmm, I wonder if “features” works at all with a touchscreen?
[16:42:41] * jthornton wanders out to the shop
[16:42:48] <Roguish> mozmck. i do not disagree at all. i am definitely a KISS kind of guy.
[16:44:59] <andypugh> Oh! Wait! The same demo in English rather than Russian:
https://www.youtube.com/watch?v=giJUiZVTXas
[17:02:59] <kwallace> And I thought it was just me.
[17:33:24] <kwallace> In reference to UI's, and I'm am completely biased, but here is a scenario that shows off PathPilot. I have a bar that is 6 inches long and I need a hole .5 inches in from each side and centered. I clamp the workpiece in the vice, change to a probe tool, click a button to probe and zero Y between the vice jaws, click a button to probe +X and zero, change to a drill tool, open the drill table and enter (.5,0),(5.5,0), set Z DROs, Post, go, then d
[17:34:22] <kwallace> Oops, Post, go, then done.
[17:43:32] <andypugh> How convincing, I just managed to get my Vismach VMC stuck on the limit switch :-)
[20:02:21] <skunkworks> http://www.machsupport.com/forum/index.php/topic,30557.new.html#new
[20:03:40] <skunkworks> pcw_home: ^
[20:09:00] <skunkworks> well I meant
http://www.machsupport.com/forum/index.php/topic,30557
[20:21:28] <PCW> yeah he called me
[20:29:24] <skunkworks> PCW, really? worth it?
[20:29:42] <skunkworks> sounds like a lot of machines...
[20:29:53] <PCW> Not sure
[20:37:17] <PCW> 4 hm2_eth cards up more that 24 hours now at 3 KHz , seems OK
[20:44:00] <CaptHindsight> PCW: just skimming the backlog, are multiple hm2_eth cards working off of one Ethernet controller now vs just 1 to 1?
[20:45:15] <PCW> yes, 4 Ethernet cards on one MAC
[20:45:33] <CaptHindsight> I just saw this: PCW> jepler: did you see that I have 4 hm2_eth cards running from a single MAC at 3 KHz?
[20:45:52] <CaptHindsight> \0/
[20:46:35] <PCW> PC --> GigE MAC --> GigE Switch --> 4 Ethernet FPGA cards
[22:42:45] <KGB-linuxcnc> 05dgarr/hallib_update 977a2d9 06linuxcnc 04. branch deleted * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=977a2d9
[22:50:28] <skunksleep> Andy might be honing in on the oword issue... Cool
[22:50:47] <Tom_itx> nice
