#emc-devel | Logs for 2007-02-17

[01:48:58] <jmkasunich_> wtf is this crap from ubuntu.... update complete system restart required? they think this is windows or something?
[01:49:38] <SWPadnos> kernel update?
[01:49:45] <jmkasunich_> yeah
[01:49:54] <jmkasunich_> except I'm running the magma kernal
[01:50:02] <SWPadnos> you don't have to restart, but of course you won't be running the updated kernel until you do ...
[01:50:14] <jmkasunich_> I don't want the updated kernel
[01:50:34] <SWPadnos> you still have the non=magma kernel installed. technically, you should be able to remove it
[01:50:35] <jmkasunich_> gotta look at grub/menu.lst and see if they changed my default
[01:50:45] <SWPadnos> I don[t know if that'll screw up anything else though
[01:50:51] <SWPadnos> yeah - I was thinking of that as well
[01:51:01] <jmkasunich_> I don't have a burning desire to experiment on my system
[01:51:22] <jmkasunich_> but that stupid "please restart" icon in the corner of the screen is gonna annoy me
[01:51:30] <SWPadnos> automatic updates are the blessing and curse of Ubuntu
[01:51:37] <jmkasunich_> agreed
[01:51:41] <SWPadnos> killall -9 update-manager
[01:51:43] <SWPadnos> (or similar)
[01:52:44] <jmkasunich_> I want most updates
[01:52:46] <jmkasunich_> just not kernel ones
[01:53:10] <SWPadnos> I wonder if it is deadly to remove the non-RT kernel package
[01:54:01] <jmkasunich_> those bastards
[01:54:05] <jmkasunich_> !\
[01:54:19] <SWPadnos> the new kernel is now default for grub, no doubt
[01:54:53] <jmkasunich_> no, they didn't do that
[01:55:11] <jmkasunich_> but the did add "quiet splash" to every fscking entry in menu.lst
[01:55:22] <SWPadnos> heh
[01:55:28] <jmkasunich_> I've only remove that gawd damned crap about ten times
[01:55:50] <jmkasunich_> _and_ changed the fscking metacomments so it wouldn't come back
[01:55:54] <SWPadnos> actually, there's probably a "linux-kernel" package that always gets modified to depend on the latest kernel - if you uninstall the empty package, you should stop getting kernel updates
[01:55:55] <jmkasunich_> but it comes back anyway!
[01:56:32] <SWPadnos> you need to put the default line and the config you want outside the "package managed" section. unless that doesn't work ;)
[01:56:55] <jmkasunich_> ## additional options to use with the default boot option, but not with the
[01:56:55] <jmkasunich_> ## alternatives
[01:56:55] <jmkasunich_> ## e.g. defoptions=vga=791 resume=/dev/hda5
[01:56:55] <jmkasunich_> # defoptions=quiet splash
[01:57:12] <jmkasunich_> I changed that line specifically so it wouldn't put the quiet/splash crap in
[01:57:20] <jmkasunich_> and it restored the line
[01:57:24] <jmkasunich_> * jmkasunich_ googles update-grub
[01:57:41] <SWPadnos> did you remove the line or make it #defoptions=
[01:58:32] <jmkasunich_> made it #defoptions=
[01:58:38] <SWPadnos> hmm
[01:59:28] <SWPadnos> http://ubuntu-tutorials.com/2006/12/29/tweaking-grub-ubuntu-510-6061-610/
[02:01:09] <SWPadnos> hmmm - that isn't too helpful, I guess
[02:01:39] <jmkasunich_> its update-grub that is doing the dirty deed
[02:01:53] <SWPadnos> https://launchpad.net/ubuntu/+source/grub/+bug/21412
[02:28:08] <jmkasunich_> it is a bug (IMO) in update-grub
[02:29:05] <jmkasunich_> if you have #defoptions = foo, it will respect your wishes
[02:29:18] <jmkasunich_> (and add foo to every kernel)
[02:29:48] <jmkasunich_> but if you have "#defoptions =" it decides that you must want quiet splash, and replaces your defoptions line
[02:30:06] <jmkasunich_> who the hell are they to override my configuration?
[02:56:03] <jmkasunich_> update-grub is a shell script, so at least you can hack on it
[02:56:35] <jmkasunich_> it makes no distinction between a menu.lst that doesn't contain "defoptions", and one that contains "#defoptions="
[02:56:51] <jmkasunich_> in either case it uses its default (which is quiet splash)
[02:57:09] <jmkasunich_> seems to me if someone explicitly puts a "#defoptions=" line, it means they don
[02:57:19] <jmkasunich_> they don't want any steenkin options
[03:02:36] <jtr> jmkasunich_: last resort, make the file read-only?
[03:02:48] <jmkasunich_> naw
[03:03:02] <jmkasunich_> update-grub is a script - I can change its defaults pretty easily
[03:03:12] <jmkasunich_> that will last at least until the next time grub is updated
[03:03:46] <jmkasunich_> or I can put "#defoptions=foo", where "foo" is some option that the kernel will ignore
[03:03:55] <jmkasunich_> (dunno what that might be)
[03:16:17] <SWPadnos> how about "nosplash"
[03:17:37] <jepler> # defoptions=resume=/dev/hda5
[03:17:40] <jepler> this is what I ended up with
[03:17:49] <jepler> if you don't use software suspend, resume= won't be used
[03:17:55] <jmkasunich_> I used panic=0
[03:18:04] <jmkasunich_> 0 is the default for panic anyway
[03:18:43] <jepler> I think that 'key=value' for unrecognized key is acceptable, and is put in the environment of 'init'
[03:19:14] <jmkasunich_> yeah, thats what I read... the trick is making sure you pick something that is indeed unrecognized by both kernel and init
[03:19:38] <jmkasunich_> nosplash is probably a good choice - if its recognized, it should do the right thing, and if not, oh well...
[03:21:21] <jmkasunich_> update-grub is just _wrong_ though... its one thing to read "comments" and use them to modify the file
[03:21:31] <jmkasunich_> but it rewrites those commands
[03:21:33] <jmkasunich_> comments
[03:21:48] <jmkasunich_> and it even erases other comments, ones with ## in front of them
[03:22:14] <jmkasunich_> (I added a ## comment explaining why the "#defoptions=panic=0" line was there, and update-grub deleted it
[03:22:45] <jmkasunich_> messing with comments violates the principle of least astonishment
[03:24:28] <jepler> jmkasunich_: I notice your rueful checkin comments -- if you install the RT kernel you can compile emc for it without booting that kernel. you may need to specify --with-realtime= though
[03:25:57] <jmkasunich_> jmkasunich_ is now known as jmkasunich
[03:43:46] <jmkasunich> I gotta figure out a better way to run the farm
[03:49:50] <skunkworks> total organic farming?
[03:53:33] <jmkasunich> with only one running at a time, build times are 5-6 minutes
[03:53:40] <jmkasunich> with all running at once, it just explodes
[04:03:46] <jmkasunich> http://pastebin.ca/359929
[04:04:07] <jmkasunich> any clue why the system keeps forgetting where libesmtp is?
[04:05:07] <jmkasunich> I run ldconfig /usr/local/lib, and it works again
[04:05:09] <jmkasunich> for a while
[04:05:30] <jmkasunich> I dunno if it just randomly dies, or if its happening when I reboot
[04:05:51] <jmkasunich> I'm sure I forget to do the ldconfig after a boot
[04:06:02] <jmkasunich> but it doesn't seem like I should have to
[04:07:48] <cradek> man ldconfig still talks about /etc/ld.so.conf, where you can list directories - however that doesn't exist on my machine
[04:08:33] <jmkasunich> I was just looking into that
[04:08:42] <cradek> that used to be the trick, not sure if it still is
[04:08:57] <jmkasunich> maybe I'll try a one-line file that lists /usr/local/lib
[04:09:31] <cradek> yeah that's what I suggest
[04:09:42] <jmkasunich> does ldconfig run at boot?
[04:10:13] <cradek> I think so
[04:11:00] <jmkasunich> ldconfig, being a user process, must be run manually
[04:11:08] <jmkasunich> from the bugs part of the manpage
[04:13:13] <jmkasunich> ld.so.conf does make ldconfig read the dir when run with no args
[04:13:37] <jmkasunich> I'm not gonna reboot to test further
[14:29:20] <jepler> jmkasunich: if it's only a problem on the ubuntu machine, why not install the esmtp package from universe?
[14:32:06] <jepler> you could also recompile the binary using libesmtp with -Wl,-rpath,/usr/local/lib in the link step
[16:44:20] <rayh> I've got libgnomeprintui2.2-0 on my system but config is looking for libgnomeprintui-2.2-0. Is that an intensional difference?
[16:56:04] <jepler> the debian package name is libgnomeprintui2.2-dev, the pkg-config package name is libgnomeprintui-2.2
[16:58:35] <rayh> Why doesn't config find it on my setup?
[16:58:51] <rayh> This is a 2.1.0 installed from the latest CD.
[16:59:01] <jepler> what does this print? pkg-config --libs libgnomeprintui-2.2
[17:00:41] <rayh> not found
[17:00:59] <jepler> did you install libgnomeprintui2.2-dev? It shold have been installed by apt-get build-dep emc2.
[17:01:30] <rayh> Nope not installed.
[17:02:00] <rayh> let me try that build dep again and see if it finds it.
[17:03:04] <rayh> Doesn't seem to "need" that package.
[17:05:24] <jepler> OK -- now I spot the error
[17:05:29] <jepler> you should be able to install that package by hand
[17:05:49] <jepler> I'll check in a fix so that after emc 2.1.1, libgnomeprintui2.2-dev is installed by build-dep
[17:06:08] <rayh> Thanks jeff.
[17:15:51] <cradek> I updated emc2-cvs-build-dep
[17:19:17] <rayh> Thanks guys.
[17:34:08] <rayh> Okay I'm curious. If latex2html isn't required, how do we get html for docs and man?
[17:37:55] <jepler> rayh: HTML documentation is not included in the .deb, so therefore latex2html is not required to build the deb
[17:38:00] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[17:38:18] <jepler> it is still required if you specify --enable-build-docs=yes or --enable-build-docs=html
[17:38:33] <rayh> I see.
[17:38:53] <rayh> And the second half. Making man pages into html?
[17:39:13] <jepler> that is also not done when building the .deb package
[17:39:26] <jepler> (or need not be, since those files are not included in the .deb)
[17:40:20] <rayh> I guess I'm thinking less legal and more proceedural.
[17:41:03] <rayh> What package or scripts are used to produce them if a user asks.
[17:41:31] <jepler> latex2html for .tex -> html and groff for man -> html
[17:41:37] <jepler> that is not changed
[17:41:48] <rayh> ah. Thanks
[17:41:55] <jepler> what is changed is that a user need not install those to rebuild the .deb package, and that's what debian/control Build-Depends: is for
[17:42:08] <rayh> Okay I get it.
[17:42:16] <rayh> Thanks for the lesson.
[17:43:19] <jepler> I explain things as well as I can...
[17:45:02] <rayh> You did great.
[20:22:45] <alex_joni> jepler: have you touched/used automake before?
[20:23:06] <jepler> alex_joni: no. it's a butt ugly pile of crap, and I advise you to stay away from it at all costs
[20:24:22] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[20:24:44] <alex_joni> jepler: lol, I can see that from the first 5 minutes around it
[20:35:29] <jmkasunich> a steaming cow patty?
[20:35:53] <jmkasunich> * jmkasunich hides from GuestXXX
[20:36:27] <jepler> uh oh there must be something wrong with my /ignore
[20:36:39] <SWPadnos> oh no
[20:36:48] <alex_joni> heh
[20:36:49] <SWPadnos> dear friends, I am confused
[20:36:51] <jepler> it's a different IP address this time
[20:36:58] <jepler> * jepler ignores it and moves on
[20:37:03] <SWPadnos> 200.180.xxx ...
[20:37:28] <jepler> ignore a whole /16? I hope that remains overkill
[20:38:14] <SWPadnos> heh. I haven't noticed whether the third byte is the same, it could be a /8
[20:38:51] <SWPadnos> jack seems to not really understand what he's doing ...
[20:38:57] <jepler> no, they're different, 200.180.186 and 200.180.89
[20:39:02] <SWPadnos> bummer
[20:39:42] <jepler> what can you do ...
[20:41:02] <alex_joni> just popping in #emc and saying one has a problem is not really usefull
[20:41:58] <jepler> jmkasunich: is there any reason *not* to provide an option to set pins to a known value on shutdown? I understand it's not a substitute for other safeguards, but it seems like a nice characteristic
[20:42:00] <SWPadnos> he also said "hi friends" - don't forget that
[20:42:17] <jmkasunich> jepler: good question
[20:42:31] <jmkasunich> the only thing I can think of is how to config it
[20:42:36] <SWPadnos> jepler, there is no reason to prevent it, but it isn't useful since you can never know the state before EMC runs, and that has to be made safe by hardware
[20:42:40] <jmkasunich> I guess there could be params
[20:42:50] <jmkasunich> pin-xx-shutdown
[20:43:04] <jepler> run a HAL file at shutdown: unlinkp parport.0.pin-01-out; setp parport.0.pin-01-out TRUE
[20:43:17] <jmkasunich> that is a far more general method
[20:43:22] <alex_joni> I like the shutdown hal file better
[20:43:24] <jmkasunich> and I thought about suggesting it to owhite
[20:43:25] <jepler> in that case It would be a feature of the emc runscript
[20:43:37] <jmkasunich> but I didn't want to have to explain it :-(
[20:43:38] <alex_joni> just like the postgui might be
[20:44:03] <alex_joni> jepler: right now the pustgui hal file is AXIS specific.. right_
[20:44:04] <alex_joni> ?
[20:44:40] <jepler> alex_joni: yes, there's not a generic way to say whether the gui is ready, and it's not needed for any gui *but* axis since only it creates HAL pins
[20:44:54] <jmkasunich> even a simple script like the pyvcp could do that... set stuff up, open a pyvcp, do a waitusr pyvcp, after the waituser, set output states, then sleep long enough to know that they have been written to the hardware, then clean up
[20:45:08] <jepler> jmkasunich: yes
[20:45:18] <jepler> so it's really not a HAL feature, but just a way of designing your HAL-using system
[20:45:23] <jmkasunich> right
[20:45:40] <jmkasunich> I'd much prefer that to having to add shutdown states to every single output driver
[20:45:49] <alex_joni> you can always spawn a halcmd out of a hal file
[20:45:51] <jmkasunich> because I'm a lazy coder
[20:46:07] <alex_joni> make it wait on some process, and run the shutdown .hal after that
[20:48:39] <jepler> http://pastebin.ca/360958
[20:48:51] <jepler> make [HAL]SHUTDOWN be a HAL script that is run after the GUI is shut down
[20:50:47] <jmkasunich> yep
[20:51:18] <jmkasunich> uh, not quite
[20:51:41] <jepler> no?
[20:51:45] <jepler> I haven't tested it ..
[20:51:49] <jmkasunich> what you wrote will use [SHUTDOWN]HALFILE as the shutdown script
[20:51:54] <jepler> oops
[20:53:21] <jmkasunich> thats a detail, the general idea is great
[20:53:45] <alex_joni> yup, I like it too
[20:57:52] <jepler> hm
[20:58:15] <alex_joni> jepler: ?
[20:58:31] <jepler> Cleanup() has both the task of cleaning up *this* emc instance *or* an old instance that didn't exit
[20:58:39] <alex_joni> right..
[20:58:46] <alex_joni> so shutdown should do that too :)
[20:58:50] <jepler> is it appropriate to run *this* emc's [HAL]SHUTDOWN when it's shutting down *another* emc session?
[20:59:09] <jepler> maybe I should only do this when "$1" != "other"
[20:59:41] <jmkasunich> I dunno...
[21:00:22] <jmkasunich> if you have multiple EMC configs for the same machine/hardware, seems like the shutdown script that renders them safe will be the same for all
[21:00:47] <alex_joni> the worst that can happen is having a halfile referring to a driver not loaded
[21:00:59] <alex_joni> but if it's an rogue emc2 it shouldn't matter much
[21:01:34] <jmkasunich> might want a -kf in there, so if there is something missing it will finish up anyway
[21:01:38] <alex_joni> although.. as usual.. people running this can seriously cause havoc (e.g. run a config with a shutdown file that triggers a selfdistruct of the machine)
[21:14:25] <jepler> that's why I'm leaning towards changing it to do nothing in the "shutting down some other emc instance" case
[21:14:41] <jmkasunich> http://www.drawblog.com/images/20070217011416651.jpg
[21:14:48] <jmkasunich> handy little tool
[21:15:04] <jmkasunich> go to drawblog.com, sketch, submit, post URL
[21:15:37] <jmkasunich> for those cases where a pic is worth a thousand words
[21:16:02] <jepler> it is a masterpiece
[21:16:17] <alex_joni> wow, nice
[21:16:20] <alex_joni> what is it?
[21:16:25] <alex_joni> :-P
[21:16:46] <jmkasunich> it would be really nice if the drawing tool would let you do text
[21:17:41] <jmkasunich> needs flash tho
[21:18:06] <skunkworks> undo is cool - but it is pretty hard to erase sections
[21:18:18] <jmkasunich> yeah, there are definitely limits
[21:18:20] <jmkasunich> but hey, its free
[21:18:25] <jmkasunich> pastebin, imagebin, and now this
[21:18:31] <jmkasunich> all handy to have
[21:18:37] <skunkworks> you can change the color to white - I guess
[21:19:06] <alex_joni> http://www.drawblog.com/images/20070217011851225.jpg
[21:19:21] <jmkasunich> showoff
[21:19:26] <alex_joni> lol
[21:19:43] <jmkasunich> I wonder what percentage of the sketches are obscene
[21:21:18] <skunkworks> <img src='http://www.drawblog.com/images/20070217012102381.jpg' usemap=#20070217012102381 border=0><MAP NAME="20070217012102381"><AREA HREF="http://www.drawblog.com" ALT="Free Sketch Hosting" SHAPE=RECT COORDS="0, 269, 80, 277"></map>
[21:21:29] <skunkworks> oopse
[21:21:58] <skunkworks> http://www.drawblog.com/images/20070217012102381.jpg
[21:23:09] <alex_joni> jmkasunich: skunkworks just posted the first obscene one
[21:23:27] <jmkasunich> alex_joni: you must have a dirty mind
[21:23:44] <jmkasunich> besides, I'm sure its far from the first