#emc-devel | Logs for 2010-03-16

[00:13:55] <jmkasunich> well crap, I had no idea the reduced cost signup ended so soon
[00:14:49] <cradek> it wasn't very reduced
[00:14:54] <jmkasunich> no
[00:15:40] <jmkasunich> the workshop doesn't have a one- or two-day pass price does it?
[00:15:50] <cradek> I don't think so
[00:16:07] <skunkworks> all or nothing..
[00:16:17] <jmkasunich> the maker's group I've gotten involved has several people interested in CNC, but I'm not sure they're $150 interested
[00:17:46] <jmkasunich> bbl, food
[01:28:30] <seb_kuzminsky> jmkasunich: does your maker group have a website? What do y'all hack on? i'm looking for something similar around here but havent found anything yet
[01:29:18] <jmkasunich> they are very new, and haven't done much yet
[01:30:05] <jmkasunich> http://www.makersalliance.org/
[01:31:25] <seb_kuzminsky> heh, nitrogen ice cream, yum!
[01:31:59] <jmkasunich> ?
[01:32:28] <seb_kuzminsky> it's one of the things listed on the "about" page
[01:34:03] <jmkasunich> did the website format just change?
[01:34:40] <jmkasunich> maybe when I loaded the homepage it was in cache, and the about page came from the server - the style is totally different, the web guy said something about remodeling
[01:37:28] <seb_kuzminsky> looks consistent to me
[02:55:46] <cradek> * cradek condescendingly pats logger_dev on the head
[02:56:12] <SWPadnos> weird
[02:56:20] <SWPadnos> I only started that once, not twice
[08:29:06] <e-jones> e-jones is now known as e-jones|away
[10:14:49] <e-jones|away> e-jones|away is now known as e-jones
[11:27:41] <jepler> SWPadnos: I think one of the things I did to try to make it auto-reconnect causes that (and doesn't make it auto-reconnect)
[13:48:24] <jepler> cradek: did my panes patch remove the toolbar, or did something else I did do that?
[13:48:37] <jepler> * jepler just noticed its absence
[13:48:37] <alex_joni> it's set to auto-hide
[13:49:31] <SWPadnos> I think you should rewrite AXIS so it's more like Photoshop. Each function should be its own small window
[13:50:12] <SWPadnos> oh, and make the default Ubuntu theme just like MacOS X, so there's only one menu bar at the top of the screen
[13:52:54] <cradek> oh hey, I have a missing toolbar too - how did I not notice that?
[13:54:49] <jepler> I dunno!
[13:57:32] <jepler> huh, probe is interestingly broken inside mdi :-/
[13:57:55] <cradek> oh? I'm sure I've used it...?
[13:58:30] <jepler> I can show you something specific it's doing weird
[13:58:36] <jepler> maybe it's probe + o call, not sure
[13:59:15] <cradek> yes, that I've heard is weirdly broken
[14:02:43] <alex_joni> SWPadnos: I think you're thinking of gimp
[14:03:04] <SWPadnos> that too
[14:03:26] <cradek> jepler: emc/rs274ngc/gcodemodule.cc:872: warning: ‘plane’ may be used uninitialized in this function
[14:03:37] <jepler> cradek: can you spot a reason the compiler is right?
[14:04:51] <cradek> no
[14:05:02] <jepler> get_attr(..., &plane) should initialize plane OR return 0
[14:06:18] <cradek> I agree
[14:06:23] <cradek> wtf, gcc
[14:28:40] <cradek> jepler: proposed for 2.4: http://timeguy.com/cradek-files/emc/0001-Better-error-messages-for-canned-cycles-on-missing-a.patch
[14:29:39] <cradek> I tested one or two of them, but not all (but I checked twice for copy-paste errors)
[14:29:45] <cradek> unfortunately, this adds strings
[14:34:38] <aystarik> jepler, get_attr may or may not assign a value to plane. Gcc has no clue about it.
[14:36:26] <aystarik> it's interesting, why it did not complain about any other get_attr use...
[14:36:43] <aystarik> int vs. double may be?
[14:37:54] <cradek> yeah that's the mystery. same for rot, x1, ...
[14:38:09] <cradek> rot is also int
[14:41:52] <aystarik> this is a different function
[14:44:27] <cradek> sure but your reasoning applies the same
[14:46:47] <aystarik> PyAPI_FUNC(int) PyArg_ParseTuple(PyObject *, const char *, ...) Py_FORMAT_PARSETUPLE(PyArg_ParseTuple, 2, 3);
[14:47:25] <aystarik> probably, Py_FORMAT_PARSETUPLE tells GCC how to look at arguments.
[14:52:07] <aystarik> jepler, you use 'double *v' to pass a pointer to int.
[14:52:35] <aystarik> GCC complains on that.
[14:52:47] <cradek> where?
[14:53:25] <aystarik> no... I'm looking at wrong function
[14:53:41] <aystarik> static bool get_attr(PyObject *o, const char *attr_name, int *v) {
[14:53:41] <aystarik> PyObject *attr = PyObject_GetAttrString(o, attr_name);
[14:53:41] <aystarik> bool result = false;
[14:53:41] <aystarik> if(attr && PyInt_CheckAndError(attr_name, attr)) {
[14:53:41] <aystarik> *v = PyInt_AsLong(attr);
[14:53:41] <aystarik> result = true;
[14:53:41] <aystarik> }
[14:53:42] <aystarik> Py_XDECREF(attr);
[14:53:42] <aystarik> return result;
[14:53:43] <aystarik> }
[14:54:30] <aystarik> probably, '*v = 0' inserted before if() will make GCC happy
[14:57:33] <aystarik> yes, it does.
[14:59:00] <SWPadnos> or an else *v=0 clause
[14:59:04] <SWPadnos> (should work)
[15:00:07] <aystarik> SWpadnos: works too
[15:01:17] <aystarik> anyway, it's better to keep compiler happy, just to not miss the case than it complains about real problem...
[15:02:09] <aystarik> s/than/when
[15:41:42] <jepler> cradek: the missing toolbar is a stupid mistake
[15:41:43] <jepler> -grid .pane -column 0 -row 0 -sticky nsew -rowspan 2
[15:41:43] <jepler> +grid .pane -column 0 -row 1 -sticky nsew -rowspan 2
[16:03:46] <jepler> Py_FORMAT_PARSETUPLE is used by a special patched version of gcc to warn about the format string of PyArg_ParseTuple. http://article.gmane.org/gmane.comp.python.python-3000.devel/9737
[16:04:09] <jepler> similarly to how gcc knows about printf format strings and can issue warnings about the types of the varargs
[16:05:55] <jepler> cradek: re your canned cycles patch -- I wish it was possible to use format strings there, so there's only one new message instead of 6. _("%s canned cycle is not possible on a machine without %s axis") -- but it doesn't look easy to do
[16:07:55] <jepler> cradek: but anyway, it looks like an improvement .. push to v2.4_branch
[16:09:20] <cradek> similarly I wish there weren't 6 copies of all that code...
[16:14:56] <CIA-2> EMC: 03cradek 07v2.4_branch * r20ccc94871cf 10/src/emc/rs274ngc/interp_cycles.cc: Better error messages for canned cycles on missing axes
[18:02:09] <skunkworks> is there a chkdsk like utillity in ubuntu?
[18:02:29] <skunkworks> heh - that was meant for the other channel
[18:04:14] <jepler> http://lmgtfy.com/?q=is+there+a+chkdsk+like+utility+in+ubuntu
[18:04:20] <jepler> :)
[18:25:54] <ries_> ries_ is now known as ries
[18:59:50] <CIA-2> EMC: 03cradek 07v2.4_branch * r2b4273166636 10/src/emc/rs274ngc/interp_find.cc: fix g53 moves that don't specify all words, while current system is rotated
[19:00:54] <cradek> jepler: thanks for noticing that shamefulness
[19:04:31] <jepler> cradek: thanks for fixing it
[19:04:31] <skunkworks> jepler: thanks :)
[19:13:06] <jepler> cradek: tests good for me too
[19:13:23] <cradek> thanks for testing
[19:13:29] <alex_joni> skunkworks: fsck it
[19:15:43] <skunkworks> doesn't work..
[19:16:09] <skunkworks> Funny - windows doesn't see it. - linux does but I cannot find anything to fix it. ;)
[19:16:16] <SWPadnos> you can't fully fsck a mounted filesystem, so you'd have to boot from a liveCD or something
[19:16:34] <SWPadnos> what file system is it?
[19:17:01] <skunkworks> I tried ntfsfix but it only seems to fix a small number of problesm
[19:17:04] <skunkworks> vista ntfs
[19:17:38] <skunkworks> I was able to copy the user data off - so that is good.
[19:17:59] <skunkworks> (owners daughters computer...) ;)
[19:18:31] <SWPadnos> so Windows (Vista or 7) doesn't even recognize the disk?
[19:18:44] <SWPadnos> is that true if you stick it in a USB enclosure also?
[19:19:03] <SWPadnos> err, move "also" in front of "true"
[19:20:33] <skunkworks> yes - that is how I started. It see the volume but cannot access it. (so I cannot run chkdsk)
[19:20:51] <skunkworks> I even booted to the recovery console - doesn't see it.
[19:21:16] <SWPadnos> bummer then
[19:21:32] <skunkworks> heh
[20:18:12] <seb_kuzminsky> rfc: http://git.highlab.com/?p=emc2.git;a=commitdiff;h=5f57865cb17182963570a95c7bfe7520cd48fa72
[20:23:09] <seb_kuzminsky> hm, i guess the last hunk of that commit (the src/Makefile one) should be its own commit and doesnt need any discussion
[20:36:01] <jepler> how many users do you think will ever remove at least one of the emc2-doc-xx debs?
[20:38:21] <seb_kuzminsky> no idea
[20:38:23] <seb_kuzminsky> me, at least
[20:38:28] <seb_kuzminsky> my french is way too rusty
[20:39:01] <seb_kuzminsky> on the computer that's dedicated to running my mill i'll probably install emc2.deb and neither of the doc debs
[20:39:26] <jepler> the package installs .desktop files that act as shortcuts to the pdf docs. it looks like those are not in the .files of the new doc packages
[20:39:39] <seb_kuzminsky> oops
[20:39:50] <seb_kuzminsky> yeah the shortcuts should move to the doc debs
[20:39:54] <jepler> consider putting gcode.html in the base package; it's very useful and tiny
[20:40:31] <jepler> (and it links to the online version for various sucky reasons anyway, so uninstalled emc2-docs doesn't diminish its usefulness in any way)
[20:40:44] <cradek> can installing the docs package change gcode.html to point to local stuff?
[20:40:56] <cradek> oh it's always to the online? I didn't know.
[20:40:59] <jepler> cradek: no, because I don't know a portable way to open to the right location inside the pdf file
[20:41:12] <cradek> I thought the htmls were installed locally
[20:41:29] <jepler> no, the package contains pdf documentation only
[20:41:35] <seb_kuzminsky> cradek: sure, the doc packages could override the gcode.html from emc2.deb
[20:41:51] <cradek> hm, ok
[20:41:57] <cradek> (hate pdf grr)
[20:42:14] <jepler> me too but there was a discussion and a consensus was reached for pdf
[20:44:40] <aystarik> m/b change next-generation into second generation? :)
[21:07:52] <alex_joni> with separate doc packages, we could have one with html docs
[23:08:20] <seb_kuzminsky> ok, i think this addresses the issues with the previous attempt: http://git.highlab.com/?p=emc2.git;a=shortlog;h=refs/heads/doc.deb
[23:08:30] <seb_kuzminsky> i'd be happy to add html doc packages too if that'd be useful
[23:08:32] <seb_kuzminsky> bbl
[23:52:15] <mozmck> Some folks are just getting a black screen with a blinking cursor when trying to boot my rtai karmic kernel. How do I troubleshoot that?
[23:55:32] <jepler> in the grub menu, remove "quiet splash" from the commandline, then photograph or transcribe whatever is on the monitor when it stops scrolling. that'll probably get you some kind of information
[23:55:36] <jepler> beyond that I don't know of a lot of tricks
[23:56:06] <mozmck> I'll see if I can't get someone to do that.
[23:56:57] <mozmck> I had one computer out of 3 or 4 I tried my kernel on that did something similar, but I think I did remove "quiet splash" and it never got far enough to show anything.
[23:57:50] <mozmck> are we going to compile for smp by default for 10.04?