#linuxcnc-devel | Logs for 2015-12-16

[07:26:25] <jthornton> I think I solved the gladevcp mystery... you can't just run gladevcp some.glade but you can run gladevcp_demo some.glade
[07:31:18] <KGB-linuxcnc> 03John Thornton 052.6 ca44da2 06linuxcnc 10configs/apps/gladevcp/helloWorld/README Docs: fix incorrect example syntax and typo * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ca44da2
[07:32:52] <KGB-linuxcnc> 03John Thornton 052.7 da52410 06linuxcnc Merge branch '2.6' into 2.7 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=da52410
[07:34:08] <KGB-linuxcnc> 03John Thornton 05master 4efd142 06linuxcnc Merge branch '2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=4efd142
[08:22:39] <linuxcnc-build> build #3028 of 4008.deb-precise-amd64 is complete: Failure [4failed shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4008.deb-precise-amd64/builds/3028 blamelist: John Thornton <jt@gnipsel.com>
[09:21:35] <JT-Shop> rut row
[09:21:56] <jepler> > WARNING: untrusted versions of the following packages will be installed!
[09:22:24] <jepler> unless you were changing something about required packages, I think this is just one of those spurious errors
[09:23:15] <JT-Shop> ok
[11:43:14] <skunkworks> zlog
[14:48:14] <micges> jepler: final patch: http://pastebin.ca/3289333
[14:59:00] <jepler> micges: is it still inconvenient for you to push it to git.linuxcnc.org or as a github pull request? If so I can go ahead and make a commit of it myself.
[14:59:20] <micges> I'll do it
[14:59:25] <jepler> OK
[14:59:38] <jepler> + if(len > sizeof(m.command) - 1) {
[14:59:39] <jepler> + PyErr_Format(PyExc_ValueError, "MDI commands limited to %d characters", sizeof(m.command) - 1);
[14:59:41] <jepler> return NULL;
[14:59:52] <jepler> I think this may get a compile warning on 64-bit, and possibly a problem at runtime
[14:59:59] <jepler> the type of this expression is size_t: sizeof(m.command) _ 1
[15:00:02] <jepler> the type of this expression is size_t: sizeof(m.command) - 1
[15:00:07] <jepler> but the type required by %d is int
[15:00:27] <jepler> for sprintf you would write %zd but I don't know the state of whether Python's PyErr_Format supports that
[15:01:07] <jepler> Modules/_multiprocessing/connection.h: PyErr_Format(PyExc_IOError, "invalid handle %zd",
[15:01:13] <jepler> python 2.7 supports %zd there
[15:01:21] <jepler> or at least they use it in python core
[15:01:51] <micges> I want to push it to 2.6 so we need to check python 2.6 too
[15:01:55] <jepler> yes, 2.6 supports %zd https://docs.python.org/2.6/c-api/string.html#c.PyString_FromFormat
[15:02:22] <jepler> hm %zu is more correct I guess
[15:13:23] <micges> fixed
[15:13:29] <jepler> thanks!
[15:13:54] <jepler> hm one more concern, let me check something first
[15:15:52] <jepler> OK, I was checking what the proper argument types for s# were
[15:16:00] <jepler> but it is const char *, int
[15:16:58] <jepler> micges: what led you to change int -> unsigned int len?
[15:17:25] <micges> compile warning
[15:17:33] <micges> probably should be size_t
[15:17:35] <jepler> at the comparison < sizeof ... ?
[15:17:38] <micges> yes
[15:18:28] <jepler> no, it can't be size_t, it has to be int (or, if PY_SSIZE_T_CLEAN is defined, Py_ssize_t) to be used in PyArg_ParseTuple https://docs.python.org/2/c-api/arg.html
[15:25:21] <micges> ugh
[15:25:29] <jepler> yeah ugh
[15:25:54] <jepler> maybe write if(unsigned(len) < sizeof...)?
[15:48:53] <jepler> .. in practice, it'll work to pass a pointer to unsigned int where a pointer to signed int is expected
[15:49:03] <jepler> so it's probably fine to leave it like that
[15:50:26] <jepler> (they have the same size and for positive values the same bit layout)
[16:04:39] <jepler> in fact passing the pointer to unsigned int instead of the pointer to int is standard C except for the strict aliasing rule, which we disable with -fno-strict-aliasing. so it's fine
[16:05:16] <jepler> > [For signed integer types, e]ach bit that is a value bit shall have the same value as the same bit in the object representation of the corresponding unsigned type
[16:06:38] <andypugh> Well, that’s a new one. When I enable the drives, the keyboard stops working.
[16:07:01] <andypugh> And plugging / replugging doesn’t bring it back.
[16:15:35] <micges> jepler: I choose unsigned(len)
[16:22:19] <jepler> OK, I am fine with that
[16:27:08] <micges> jepler: are you doing python + c in dayjob?
[16:27:26] <jepler> micges: yes
[17:03:11] <Skunkworks> Andy - EMI sucks
[17:05:22] <Skunkworks> I was running our cobbled together router last night and found out unplugging the running spindle (a rotozip) causes noise and lost steps
[17:11:08] <andypugh> Not sure it is EMI.
[17:11:42] <andypugh> Though I can’t turn the drive back of again at the moment as the GUI and PSU are not linked.
[17:14:26] <KGB-linuxcnc> 03Michael Geszkiewicz 052.6 ff59490 06linuxcnc 10src/emc/usr_intf/axis/extensions/emcmodule.cc emcmodule: Fix incorrect memory access by PyArg_ParseTuple * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ff59490
[20:34:46] <jepler> thank you micges!
[22:31:02] <mozmck> heh, new install of jessie uses LibreOffice to open .sh files.
[22:34:18] <mozmck> huh, can't run the vbox guest additions installer
[22:38:30] <mozmck> argh! user created at install is not in sudoers file!
[22:40:19] <cradek> sadly, it took a lot of futzing on our wheezy cd to not ask for root password at the installer, have working sudo, have update manager automatically checking for updates, etc etc
[22:40:53] <mozmck> I bet!
[22:41:07] <cradek> by default, debian is really still a multi-user unix setup with users and root expected to be separate people
[22:41:21] <mozmck> Do we want automatic update checking?
[22:41:39] <cradek> I sure want users to have that
[22:41:48] <cradek> the thing should pop up and say there are updates
[22:41:58] <mozmck> by default, debian is really still geared toward sys-admins and not general desktop users.
[22:42:12] <cradek> however, it shouldn't offer an upgrade to a different distribution
[22:42:34] <mozmck> Ok. I think a lot of people don't even have their machine control internet connected.
[22:42:49] <mozmck> Does the debian update checker do that?
[22:43:45] <cradek> http://timeguy.com/gitweb?p=live-images.git;a=commitdiff;h=4bd9ac90d8dd5c04ad6ca5c64a08ed0ffb49fe57
[22:44:11] <cradek> preseeding passwd/user-default-groups like this lets you put the installed user in the sudo group
[22:44:34] <mozmck> I pulled that the other day but haven't had time to look at it yet. I was just installing jessie w/Mate to play with it more.
[22:44:53] <cradek> what do you think of it?
[22:45:28] <mozmck> It's like gnome used to be so far as I can tell in a few minutes :-)
[22:45:37] <cradek> that's good
[22:45:53] <cradek> I predict it's just like gnome2 but subtly brokener
[22:46:02] <mozmck> I've gotten used to some things in xfce now. Like having "Open in terminal" by default in the right-click menu in the file manager.
[22:46:29] <mozmck> why brokener? It is gnome2 +
[22:46:35] <cradek> because of the +
[22:46:39] <mozmck> ah
[22:46:43] <cradek> :-)
[22:46:52] <mozmck> I wonder if it still uses mostly gtk2?
[22:46:57] <cradek> it's just a feeling I get. maybe unjustified.
[22:48:04] <mozmck> one thing I really like is the plugin for Nautilus/Nemo that puts a small terminal window in the top of the files section, and it follows the current directory.
[22:52:37] <cradek> goodnight mozmck