#linuxcnc-devel | Logs for 2014-11-26

[00:13:39] <cmorley> cradek: Thanks for the report on sim_parport - i take it it's ok though? I probably should change the invert pins to ! instead of ^ 1
[00:13:57] <cmorley> easier to see what I was intending.
[00:51:58] <KGB-linuxcnc> 03Chris Morley 052.7 06c96cf 06linuxcnc 10src/hal/components/sim_parport.comp HAL -improve readability of the invert code * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=06c96cf
[08:38:09] <cradek> cmorley: yes, turns out it was fine because that type changed to bool a while back, but I like your change anyway
[09:02:12] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff 88311eb 06linuxcnc 10configs/sim/axis/moveoff/moveoff_gui.tcl 10src/hal/components/moveoff.comp moveoff.comp: support font spec, wider +/- * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=88311eb
[12:05:59] <skunkworks> I bet with deweys moveoff you could have it go back to the original location.. It could set the FO or MV or something to zero until the offsets where removed..
[12:11:56] <archivist> would it know which order they had to be done in so as not to gouge the work and fixtures
[12:54:33] <skunkworks> maybe set the order.. or the last axis to move..
[12:58:38] <seb_kuzminsky> the comp gets a sequence of offset-changes, it could record that and then play it back in reverse order
[13:00:39] <archivist> could/should :)
[13:06:42] <skunkworks> bunch of arm chair quarterbacks..
[13:15:05] <kwallace> I was thinking of posting this: http://www.softwarebyrob.com/2006/10/31/nine-things-developers-want-more-than-money/
[14:18:06] <Connor> Boy, Jog while paused is getting lots of attention on the email list...
[14:19:57] <seb_kuzminsky> Connor: it's a popular topic, for sure
[14:21:12] <Connor> Finding the correct solution.. sounds like the hard part.
[14:21:45] <Connor> all I can say on the subject is, what EVER the solution.. it should be simple to implement..
[14:21:53] <cradek> haha
[14:22:22] <cradek> it's fairly immune to simple solutions
[14:22:58] <seb_kuzminsky> dewey's solution, while incomplete and limited in certain ways, is pretty simple
[14:23:19] <Connor> Yea, but, it doesn't update axis..
[14:23:28] <seb_kuzminsky> that's one of the limitations
[14:23:30] <cradek> I'm afraid my "here are the complexities, let's think about them and make a real design" attempt was just seen as obstructionism
[14:23:56] <seb_kuzminsky> cradek: yeah, i noticed that too
[14:23:59] <seb_kuzminsky> that's sad
[14:24:06] <cradek> very frustrating
[14:24:40] <Connor> Michael Haberler had something to make it work.. not sure how long ago that was.. what was wrong with that approach ? 3
[14:25:01] <cradek> that's been discussed at length recently
[14:25:06] <cradek> in irc though - not the list
[14:25:35] <Connor> Okay. I missed it.. I have full time job, so don't get to read everything on IRC all the time.. just bits here and there.
[14:25:46] <cradek> me too :-)
[14:26:06] <Connor> That comment was not to imply that others don't have full time jobs BTW.
[14:26:52] <seb_kuzminsky> heh yeah
[14:27:18] <Connor> okay, so what was the gist of the irc talk ?
[14:27:46] <cradek> brb
[14:33:14] <seb_kuzminsky> Connor: i think this is the discussion cradek's talking about: http://psha.org.ru/irc/%23emc-devel/2014-11-21.html#20:38:05
[14:41:19] <Connor> okay.. I picked up that last bit of that conversation..
[14:44:20] <seb_kuzminsky> dgarr (when you read back): i really like moveoff_demo2
[14:45:08] <seb_kuzminsky> it's got all the features/behaviors i've imagined for the solution you're implementing
[14:45:12] <seb_kuzminsky> nice work
[14:46:08] <seb_kuzminsky> however, something weird happened when i un-checked "Enable Offsets" while there were offsets active
[14:46:39] <seb_kuzminsky> halscope showes the offsets getting un-done right away, but the main offsetting window got confused
[14:47:17] <seb_kuzminsky> it said "Removing offsets" in yellow at the bottom, and the box to the right of the Y offset buttons started counting down, and just kept going
[14:47:26] <seb_kuzminsky> it's at -5.4075 now and still going
[14:47:47] <andypugh> Is there an option to hook up jogwheels and axis selection switches? That could make it feel a lot more like a true JWP on machines with such devices.
[14:49:56] <seb_kuzminsky> andypugh: no, not yet
[14:49:58] <cradek> it seems like you could use the component in lots of ways, not just with the special sub-gui
[14:50:55] <Connor> Okay, so, why doesn't it update the gui? or is that a whole other ball of wax? :)
[14:50:56] <cradek> the problem with making a sample gui implementation is that's exactly/only what people see
[14:51:34] <cradek> then you get THIS JWP IS USELESS THE BUTTONS ARE TOO SMALL and other similar nonsense
[14:51:40] <andypugh> I think that this might be a problem with the Machinekit demo config, I think that keyboard jogging is also possible there.
[14:51:55] <cradek> that would be (big) news to me
[14:52:20] <cradek> at least if you mean the normal guis' normal jogs
[14:52:57] <andypugh> I think (if I was making the component) that I would want to add “shadow” versions of the halui jog inputs to be connected to the same signals.
[14:53:09] <Connor> I'm going to want to jog from multiple places.. remote pendant (xhc-hb04), local MPG + axis selector, and keyboard, and GUI
[14:53:39] <cradek> oh that could be cool - it could have the same wheel interface as everything else: enable/scale/counts
[14:53:58] <cradek> with the minor added twist that when you disable it, it goes back to zero
[14:55:40] <cradek> ... is that what you mean?
[14:57:47] <andypugh> Yes.
[14:58:22] <andypugh> For a typical lathe that would look almost exactly like what is being asked for.
[14:59:00] <andypugh> (Spindle over-ride to zero stops the spindle)
[15:00:04] <cradek> I thought we could turn spindle on and off while paused now (with on/off, not spindle override) but it didn't work for me the other day (in AXIS)
[15:00:16] <cradek> I need to get my shit together and start filing bug reports
[15:03:06] <cradek> andypugh: I can't decide if that would be best added directly to moveoff.comp, or if another component should intervene (scale/counts => x_offset_in et al)
[15:03:32] <cradek> I bet directly is better
[15:03:48] <cradek> I like the idea of having it be as much like halui as possible
[15:04:18] <andypugh> I think directly.
[15:04:31] <andypugh> I assume that moveff is in two parts? userland GUI and realtime?
[15:05:29] <cradek> appears to me the functionality is in a (realtime) comp, and then there's a gui that's made specially to twiddle the comp's inputs
[15:05:41] <cradek> was just looking at http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=blob;f=src/hal/components/moveoff.comp;h=baa3f049b672a1543e67722d36166bc961d2d322;hb=88311eb
[15:08:05] <cradek> he has really thought through (warned about in the docs) the corner cases
[15:08:11] <cradek> easier to read: http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=blob_plain;f=src/hal/components/moveoff.comp;hb=88311eb
[15:13:30] <andypugh> It does seem very explitly xyz rather than 123..N
[15:13:40] <seb_kuzminsky> it's cool that you can do either "apply a fixed offset from the ini file" or "apply a dynamic offset from a userland jogging application"
[15:14:07] <cradek> andypugh: the pins are named xyz but you could splice them into any of your actual joints
[15:14:43] <cradek> sure could be 0..8 just as easily
[15:15:12] <andypugh> I am just thinking that extending to 9 or more joints would be painful, and comp does understand variable pin numbers.
[15:16:19] <andypugh> A lot of it looks like it could be a loop-over-joints.
[15:17:31] <andypugh> pin in float #-pos [personality:N] etc..
[15:42:40] <andypugh> The do …. while(0) construct in a #define, I assume that can’t be a loop with an index, as the index isn’t allocated?
[15:45:36] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff 8527fed 06linuxcnc 10configs/sim/axis/moveoff/moveoff_gui.tcl 10src/hal/components/moveoff.comp moveoff.comp: zero lim3() outputs after use * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8527fed
[15:45:36] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff a0522d5 06linuxcnc 10configs/sim/axis/moveoff/moveoff_demo2.ini 10configs/sim/axis/moveoff/moveoff_gui.tcl moveoff.comp: handle parameters in demo gui * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a0522d5
[16:46:02] <andypugh> 9-axis moveoff, and shorter: http://pastebin.com/FW5W1nVW
[16:48:10] <cradek> awesome, I bet dgarr would welcome that - you probably know the personality stuff better than anyone
[16:50:24] <andypugh> Well, I did finish that then notice that moveoff count=3 gives you 9 axes…
[16:51:03] <andypugh> (except that there is a subtle bug where the instances have separate data, but a shared state-machine index )
[16:54:20] <cradek> http://linuxcnc.org/index.php/english/forum/9-installing-linuxcnc/28606-debian-wheezy-26n-install-on-gx620
[16:56:34] <andypugh> And he’s a Gold Boarder..
[17:51:27] <dgarr> seb_kuzminsky: [20:26:12] <seb_kuzminsky> however, something weird -- what os and kernel was that? i have reports of similar problem on 2.6.32-122-rtai
[17:52:18] <dgarr> i have tested on 3.2.0-4-rt (rtpreempt,uspace,debian,wheezy), and 3.4.9-rtai-685-pae 3.4.55-4linuxcnc on debian wheezy both ok
[17:53:03] <dgarr> and uspace/sim on ubutntu quantal 3.5.0-19-generic which i develop on
[17:53:15] <dgarr> i will read back, holiday days now
[17:54:26] <andypugh> Rats! I missed him!
[18:04:23] <seb_kuzminsky> it was a drive-by dgarring
[18:06:33] <JT-Shop> lol
[18:13:54] <andypugh> There is a variable-allocation before variable-declaration quirk in the moveoff component that might upset some compiler flags more than others.
[18:14:28] <andypugh> FUNCTION(_) {
[18:14:28] <andypugh> theperiod = period;
[18:14:30] <andypugh> double lout,old_in,old_v;
[18:14:31] <andypugh> bool enable = move_enable && is_on && is_paused;
[18:14:31] <andypugh> bool move = 0;
[18:15:04] <andypugh> (theperiod is a double with global scope)
[18:51:07] <seb_kuzminsky> cradek: hey neat, weren't we just talking about a scale like this: http://www.watzlavick.com/robert/rocket/rocket1/photos/dsc_0081m.jpg
[18:51:56] <cradek> I ordered that one from enco with 10ths/100ths of inches on one side and mm/halfmm on the other side
[18:52:01] <cradek> and yay, it's exactly what I always wanted
[18:52:08] <cradek> what is that thing?
[19:09:42] <jepler> andypugh: I thought we enabled that C99 feature (declaration after statement) everywhere. not true?
[19:31:54] <kwallace> cradek, in case you had not found this yet : http://www.watzlavick.com/robert/rocket/rocket1/photos/
[20:33:15] <skunkworks> looks like the lathe powered up...
[20:33:27] <skunkworks> darn/yay
[20:48:09] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff bb40998 06linuxcnc 10(7 files in 2 dirs) moveoff.comp: refactor for comp personality use * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=bb40998
[21:08:43] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/moveoff 11aea19 06linuxcnc 10src/hal/components/moveoff.comp moveoff.comp: print all items when showing Error * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=11aea19
[21:13:38] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/2.7/docs-reorg 67c7f5a 06linuxcnc 10(81 files in 9 dirs) docs: reorg Getting Started (English) * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=67c7f5a
[21:13:39] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/2.7/docs-reorg dd78510 06linuxcnc 10(77 files in 8 dirs) docs: reorg Getting Started (French) * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=dd78510
[21:13:39] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/2.7/docs-reorg c6f77d8 06linuxcnc 10(19 files in 5 dirs) docs: reorg Getting Started (Spanish) * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c6f77d8
[23:18:59] <kwallace> skunkworks, I guess you need to find another machine to work on.