#linuxcnc-devel | Logs for 2014-01-14

[05:56:03] <KGB-linuxcnc> 03John Thornton 05v2.5_branch ab7e013 06linuxcnc 10docs/src/hal/rtcomps.txt Docs: add info on how to calculate base thread for software encoder * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ab7e013
[10:16:56] <seb_kuzminsky> zultron: if the build system detects a xenomai kernel, it should build both the xenomai and xenomai-kernel flavors, right?
[10:33:51] <zultron> Hi seb_kuzminsky, I hope so. :)
[10:34:23] <zultron> The intended behavior would be if the xenomai-devel pkg is present, build xenomai; if that and a xenomai kernel headers pkg are present, build xenomai-kernel.
[10:34:52] <zultron> (And if multiple xenomai kernel headers pkgs are present, build xeno-kernel modules for each)
[10:40:39] <seb_kuzminsky> ok cool
[11:21:21] <dgarr> cradek: any objections to this patch for touchy: http://www.panix.com/~dgarrett/stuff/0001-touchy-allow-alternate-gladefile.patch
[11:31:04] <cradek> the idea, no problem - but can you just use inifile.find(...)? using findall and then discarding all but the first seems weird unless I am missing something
[11:31:22] <cradek> also, is the second hunk unrelated?
[11:33:39] <cradek> is the second hunk a bugfix? you moved the last line inside the loop, and that seems wrong at first glance
[11:57:39] <dgarr> the second hunk was just editiig tab damage, i've restored that and fixed up findall in the patch. touchy.py seems to have a mixed set of tabs and indents ?
[11:58:51] <cradek> cool, can I see this version?
[11:58:56] <dgarr> using an alternate glade file for gui small enough to fit 640x480: http://www.panix.com/~dgarrett/stuff/touchy_small/
[11:59:17] <cradek> yeah mixed tabs and indents are a scourge but fixing them is worse
[11:59:18] <dgarr> patch is in the same place: http://www.panix.com/~dgarrett/stuff/0001-touchy-allow-alternate-gladefile.patch
[11:59:53] <cradek> I am surprised to learn that it won't work on 640x480 as-is
[12:00:49] <dgarr> have you tried 640x480? i couldn't get it to fit with touchy.glade
[12:01:04] <cradek> no I admit I haven't actually tried it
[12:04:58] <dgarr> with vanilla touchy.glade and all fonts==6, i get $ wmctrl -lG |grep Touchy
[12:04:58] <dgarr> 0x05000006 2 565 78 620 675 u51 LinuxCNC/Touchy
[12:07:20] <cradek> I think you are opening the inifile twice and perhaps leaking at least a fd, maybe whole python objects (I'm not sure about the scoping rules)? I think the fix would be to move the inifile opening from line 176 above where you need it to get the gladefile, then use self.ini.find
[12:08:49] <cradek> that line 176 block is in a weird place
[12:09:17] <cradek> ugh, it gets opened again later for reading UNITS
[12:09:49] <dgarr> yes it seemed it already opened more than once
[12:10:36] <cradek> could you fix that in commit #1 (also moving it above where you need it) and then add your glade stuff in commit #2?
[12:11:48] <cradek> postgui handling also does it
[12:12:01] <cradek> I can fix this (later today) if you don't want to
[12:12:22] <cradek> the embed tabs stuff does the right thing
[12:14:10] <cradek> meanwhile, I've failed to get a vncserver running at 640x480 with a working fontpath, and I'm going to lunch
[12:43:35] <dgarr> cradek: http://www.panix.com/~dgarrett/stuff/touchy_alternate_gladefile.mbox
[12:43:55] <dgarr> (two patches)
[13:14:15] <cradek> inifile=self.linuxcnc.emc.ini(sys.argv[2])
[13:14:32] <cradek> I think you missed a removal of this line around 817
[13:14:38] <cradek> still digesting the rest
[13:18:15] <cradek> also around line 254?
[13:20:33] <memleak> working on xenomai for cubieboard2 (dual core allwinner A20) kernel 3.4.76 :)
[13:22:21] <dgarr> cradek: mbox revised, i've just used wmctrl or wminfo to measure size and when small enough tested by modifying display (xubuntu and xfce)
[13:26:27] <cradek> dgarr: this looks excellent now, thanks for cleaning it up
[13:27:01] <cradek> I see another problem (assumptions about argv[2] in postgui loading) but it's getting really far from the changes you want to make, and not a dependency at all
[13:28:07] <cradek> dgarr: I'm curious - what changes do you make to the glade to let it get smaller?
[13:29:48] <dgarr> quite a bit, this patch follows the two in the mbox (work in progress): http://www.panix.com/~dgarrett/stuff/wip.patch
[13:30:28] <dgarr> bbl, please post if you find vanilla glade file can show all at 640x480
[13:30:43] <cradek> the height of the status screen is what's keeping me from sizing smaller
[13:30:54] <cradek> um, it shows all 55 pockets, that can't be right
[13:31:06] <cradek> they all have T00 in them, haha
[13:31:36] <cradek> I'm at 614x503 without changing anything except fonts with the font chooser
[13:33:26] <cradek> oh it was because my tool table didn't exist
[13:36:26] <cradek> creating a tool table lets me get to 640x480 easily: http://timeguy.com/cradek-files/emc/touchy-640x480.png
[13:36:47] <cradek> without even resorting to tiny/ugly fonts
[13:36:59] <cradek> I wonder why it's different for you
[13:57:08] <cradek> having more axes than XYZ makes it taller I bet
[15:01:13] <andypugh> Is it bad form to use the result of a comparision in a calculation?
[15:01:32] <andypugh> ie, use "true" and "false" as actual numbers.
[15:03:13] <seb_kuzminsky> yes, i think so
[15:03:28] <seb_kuzminsky> booleans are not integers, and treating them as such is confusing to my feeble brain
[15:04:18] <andypugh> So, to take a not very random example, if (a % b == (b > 1)) would be less preferred than if (a % b == (b > 1 ? 1 : 0))
[15:05:41] <seb_kuzminsky> i'd prefer "if (b == 0) { do_one_thing(); } else { do_another_thing(); }", but i understand that may not be a universal preference
[15:05:45] <andypugh> Alternatives are an extra constant, or a third clause in the if-else
[15:06:25] <seb_kuzminsky> both of the ways you wrote look foreign and too-clever for me, but then you're foreign and clever so maybe that's ok
[15:07:43] <andypugh> It's the resolver "index" code http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=blob;f=src/hal/drivers/mesa-hostmot2/resolver.c;h=3fca0a16f6eeeea30ad90098e81f5e225a0b7b96;hb=9a4a626bcee4f65071b98cc448e83fccd68356a1#l325
[15:08:27] <andypugh> It's already an 8-line "if"
[15:09:49] <seb_kuzminsky> i dont envy you the task of finding the clearest way to express that... expression
[15:11:35] <andypugh> I am not proud of that code, and it took a lot of fiddling to get it nearly right. Except in doing so I got it really wrong.
[15:12:43] <andypugh> (Because N % 1 == 1 is unlikely...)
[15:19:42] <cradek> (b? do_another_thing: do_one_thing)();
[15:24:49] <andypugh> It's a bit if a pain to get this compiled and running on a machine with an actual resolver. But I guess I have no real choice.
[15:26:27] <andypugh> I think the resolver machine is running mah's "Jog while paused" trial, and it might have compiled it from my shared git which is currently on a different branch. Oh joy :-)
[16:37:05] <andypugh> What a complete pain in the behind. The PC in the (freezing cold) garage has lost the network manager applet.
[17:04:51] <dgarr> cradek your picture shows a Mofit theme ? is that in your distro (what distro?), i don't see it in my xubuntu / quantal default install
[17:43:53] <KGB-linuxcnc> 03Dewey Garrett 05master 53410b6 06linuxcnc 10src/emc/usr_intf/touchy/touchy.py touchy: open inifile once only * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=53410b6
[17:43:53] <KGB-linuxcnc> 03Dewey Garrett 05master 469c777 06linuxcnc 10src/emc/usr_intf/touchy/touchy.py touchy: allow alternate gladefile * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=469c777
[18:25:55] <KGB-linuxcnc> 03Andy Pugh 05master b0b7e47 06linuxcnc 10src/hal/drivers/mesa-hostmot2/resolver.c Fix resolver indexing properly. * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=b0b7e47
[18:30:55] <seb_kuzminsky> awesome, actual commits that actually make things better, thanks guys :-)
[18:31:29] <andypugh> How do I un-set rip-environment?
[18:32:27] <andypugh> I guess a new terminal should do it?
[18:32:58] <andypugh> In which case I need to switch my installed version back to a JA4 version, as my config has been converted
[18:49:05] <seb_kuzminsky> yeah, new shell is the only way i know to un-rip your environment
[18:58:02] <andypugh> Touchy in 2560x1440 is a bit overwhelming.
[18:58:50] <andypugh> But that means I have my system back to working after disabling it for the resolver fix.
[18:59:26] <andypugh> Any idea where to look for this kernel compile problem bufp.c:(.text.unlikely+0x209c): undefined reference to `dma_alloc_noncacheable'
[19:00:03] <andypugh> (there is actually no file called bufp.c, but there is a bufp.o.
[19:00:12] <andypugh> Does "make" make up file names?
[19:00:35] <andypugh> Perhaps a submakefile can assembe multiple .c files into one .o?
[19:06:04] <KGB-linuxcnc> 03Sebastian Kuzminsky 05remove-obsolete-hook e3eca06 06linuxcnc 04debian/D99kernel-img.conf deb: remove unused pbuilder hook file * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=e3eca06
[19:06:24] <seb_kuzminsky> andypugh: no, each .o is compiled form one .c
[19:06:51] <seb_kuzminsky> you can combine multiple .o files, but they become .a or .so files (aka static libraries and shared libraries)
[19:07:19] <andypugh> The file turns out to exist, but neither Eclipse nor find found it, until I started looking from somewhere else.
[19:07:23] <andypugh> Very strange!
[19:07:26] <seb_kuzminsky> maybe your bufp.c is created by the build process, like our build proces makes .c files from .comps?
[19:07:30] <seb_kuzminsky> oh
[19:07:32] <seb_kuzminsky> heh
[19:07:58] <andypugh> Ah, he's a symilink!
[20:38:37] <andypugh> Is this actually valid C? The compiler doesnt seem to think so:
[20:38:45] <andypugh> http://pastebin.com/dWQ6zY8s
[20:39:06] <andypugh> (part of Xenomai, so maybe just being super-clever)
[20:53:44] <cradek> dgarr: it's debian, but any simple theme should work - I just picked that one because it's easy to tell which buttons are in vs. out, something that a lot of fancier themes do badly
[20:54:29] <cradek> I don't think the theme affected it much
[20:54:51] <cradek> I think "redmond" and "mist" were the good themes for usable controls in ubuntu/gnome
[21:02:34] <memleak> source /etc/profile should undo at least some of the rip-environment
[21:02:49] <memleak> that is...
[21:02:53] <memleak> $ source /etc/profile
[21:03:29] <andypugh> It was easy enough to open a different terminal style.
[21:05:31] <andypugh> (I am working at a Mac, and that has a number of different colour / style terminals. This sounds daft, but it makes it easier to keep track of which is which remote ssh machine, and they each have their own startup commands set.)
[21:06:34] <cradek> that sounds daft
[21:06:39] <cradek> :-)
[21:07:33] <andypugh> Ah, wait a minute, those just do auto log-in to various machines. And it is the remote end that is configured to automatically rip-environment in a .ssh thing. Which normally saves mistakes, but is hard to circumvent when I don't want that :-)
[21:09:06] <andypugh> Really should be asleep. Night all
[22:24:04] <KGB-linuxcnc> 03Dewey Garrett 05master 187b407 06linuxcnc 03configs/sim/touchy/ngcgui/lathe.tbl 03configs/sim/touchy/ngcgui/sim.tbl 03configs/sim/touchy/simpockets.tbl configs: add missing symlinks for tbl files * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=187b407
[22:57:06] <KGB-linuxcnc> 03Sebastian Kuzminsky 05unversioned-tcltk-deps de8e545 06linuxcnc 10debian/configure 10debian/control.in deb: make dependency on tcl/tk unversioned * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=de8e545