#emc-devel | Logs for 2006-06-05

Back
[01:08:39] <cradek> new EMC 2.0.1 packages in the ubuntu repository
[01:53:49] <jepler> someone should announce 2.0.1 on the website
[01:54:11] <jmkasunich> I already nominated chris to do that
[01:54:24] <cradek> I think I'm in turn nominating alex
[01:54:42] <jmkasunich> oh, you mean the linuxcnc site?
[01:54:48] <jmkasunich> yeah, thats definitely alex
[01:54:51] <jmkasunich> or maybe SWPadnos
[01:55:01] <jmkasunich> * jmkasunich doesn't do joomla
[01:55:09] <jmkasunich> I was talking about the SF main page
[01:55:37] <jepler> I meant the linuxcnc.org website
[01:55:40] <cradek> cdrecord: WARNING: Data may not fit on current disk.
[01:55:40] <cradek> cdrecord: Notice: Use -overburn option to write more than the official disk capacity.
[01:55:43] <cradek> eh??
[01:55:49] <jmkasunich> dapper?
[01:55:55] <cradek> yeah
[01:56:01] <jmkasunich> how big is it?
[01:56:20] <cradek> 731744256
[01:56:38] <SWPadnos> err - I
[01:56:40] <jmkasunich> that is kinda big
[01:56:50] <cradek> I agree
[01:56:51] <jmkasunich> I thought a CD was 700 and "a few" meg
[01:56:53] <jepler> that's the same size as the version I got on bittorrent
[01:56:55] <SWPadnos> err - I'm not exactly sure how to go about updating the linuxcnc.org site, to be honest
[01:56:59] <cradek> but the md5sum is right
[01:57:07] <jmkasunich> dodge, duck
[01:57:13] <jepler> I burned it with nautilus-cd-burner or whatever it's called, and got no warning.
[01:57:31] <cradek> jepler: that doesn't work with my old burner
[01:57:33] <jmkasunich> probably because nautilus is the "user friendly" burner
[01:57:52] <cradek> SWPadnos: I know it has something to do with those buttons that have rounded corners
[01:59:10] <cradek> crap, I can't burn it
[01:59:33] <jepler> cradek: remind me tomorrow morning that I could have brought you a copy, but forgot it at home instead.
[01:59:41] <cradek> jepler: ok
[01:59:47] <jmkasunich> ;-)
[02:00:34] <cradek> if it were a *real* unix, I could install from dat tape
[02:01:14] <jepler> cradek: I know you're just trolling, but the fact that you can't is more about the limitations of the IBM PC BIOS than anything else.
[02:01:27] <jepler> (maybe I'm mistaken and there is a way for SCSI to load a boot block from tape)
[02:01:35] <cradek> sure, can't boot from it, but it could install from it with a floppy
[02:01:47] <jepler> anyway, I challenge you to produce an ubuntu live cd
[02:01:47] <cradek> actually I'm pretty sure freebsd can (could?)
[02:01:49] <jepler> er, live tape
[02:01:52] <cradek> haha
[02:02:17] <cradek> http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/install-media.html
[02:02:36] <cradek> *ahem* and it's in text mode of course
[02:02:50] <jmkasunich> I thought it was kinda cool that I could install freebsd with 3 floppies and a net connection
[02:03:30] <cradek> jmkasunich: it used to be with laptops that's often all you had
[02:03:43] <cradek> jmkasunich: actually redhat could do it too
[02:03:51] <cradek> I think they lost that capability now in fedora
[02:05:17] <jepler> I just don't run into many machines without a cdrom drive but with a 'net connection fast enough to download 1GB of RPMs
[02:05:38] <cradek> heh, surely that's another issue
[02:05:39] <jepler> if you're only going to download 10 megs and you have the equivalent of an early 1990s university's network, that's one thing
[02:05:39] <jmkasunich> the compile farm is probably one of the few
[02:06:13] <cradek> but with the CD/DVD you have to download them whether you want them or not. surely a real network install is superior for that reason
[02:06:18] <jmkasunich> it has a shared floppy for all 8 slots, the only way to get a cdrom is to pop the top and plug into a slot's IDE connector
[02:06:56] <jmkasunich> cradek, until you do the 2nd install, the burned CD is much faster for multiples
[02:07:17] <cradek> besides any reasonable OS install is only a couple hundred meg right? that's 10 minutes download time
[02:07:26] <cradek> (now I'm trolling)
[02:11:08] <jmkasunich> hmm, the dithered pwm isn't as bad as I thought
[02:11:42] <jmkasunich> it has issues only when making a dramatic change in duty cycle, _and_ the new duty cycle is very close to either 0 or 1
[02:11:59] <jepler> "breezy" hosted in vmware took 2:30 "real" to rebuild the emc 2.0.1 deb
[02:19:40] <jmkasunich> heh! fixed it
[02:20:26] <jepler> with ccache enabled, the second './configure --with-rtai=/usr/realtime-2.6.12-magma && make' took 1:12 real
[02:20:48] <jepler> I'm getting a lower hit rate than I expected from ccache
[02:23:57] <jepler> jmkasunich: when you come back I might want to talk about using this vmware stuff as farm slots
[02:24:14] <jmkasunich> I'd like to do that
[02:24:40] <jmkasunich> I tried once with qemu and wasted an entire weekend fskcing around
[02:24:48] <jmkasunich> but if I vmware is faster....
[02:24:52] <jepler> I've already installed two guests and built emc2 on one of them
[02:24:53] <jepler> since 4PM
[02:24:59] <jepler> with 3 hours out for dinner with family
[02:25:01] <jmkasunich> I have a dual P3-600 box I can dedicate to it
[02:25:11] <jepler> anyway .. next week we can talk about it
[02:25:12] <jmkasunich> or maybe I can run it on my main box (rather not tho)
[02:25:38] <jmkasunich> how fast is your hardware?
[02:25:46] <jepler> amd64 1.8GHz
[02:26:13] <jmkasunich> so still several times faster than my P3s, and probably a good bit faster than this box
[02:26:43] <jmkasunich> this box = sempron 2200+, I think thats around 1.6G
[02:27:23] <jmkasunich> feel free to mail me, I might be able to read mail in the evenings
[02:27:34] <jmkasunich> I'll be back tuesday night (I hope)
[02:27:38] <jepler> see you
[02:27:43] <jepler> don't let me keep you
[02:27:45] <jmkasunich> goodnight
[02:32:07] <jepler> an idle guest OS seems to take about 5% CPU, so you can't run all that many of 'em
[02:42:13] <jepler> but you can 'vmware-cmd suspend' and 'vmware-cmd restart' them so they're not going all the time
[02:49:46] <jepler> (from the host)
[02:50:03] <jepler> vmware server is pretty neat, clearly a whole different realm of product than qemu
[18:07:39] <fenn> if breezy wants to automatically update to dapper and that breaks realtime, we should probably include a bugfix for that in the .deb
[18:08:26] <SWPadnos> it asks you if you want to upgrade to the new version, which most people will answer "yes"
[18:22:01] <jepler> breezy asks you to update to dapper? that's fucked up.
[18:22:18] <jepler> why'd we pick ubuntu again?
[18:22:41] <Roguish> yes it does. and yes it is.
[18:23:04] <Roguish> "just say no"
[18:23:24] <jepler> I hate everyone.
[18:25:19] <cradek> ugh
[18:25:31] <cradek> I didn't see that behaivor yesterday when I made the new emc2 packages
[18:25:55] <cradek> dammit
[18:30:37] <Lerneaen_Hydra> what? AFAIR it just said, there is a new version of ubuntu available, under the name of "dapper drake" or something like that
[18:34:13] <SWPadnos> the update manager lets you know that there's a whole new version of the OS, not just updates to the stuff you already have
[18:35:23] <cradek> until today, the update manager just had bug and security fixes in it, so it was ok to let it do its thing
[18:35:35] <cradek> (I haven't seen it yet so I can't say too much)
[18:36:55] <SkunkWorks> It is an extra button - upgrade to dapper.
[18:37:20] <cradek> ah ok
[18:40:47] <SkunkWorks> http://www.electronicsam.com/images/KandT/upgrade.png
[18:41:51] <cradek> SkunkWorks: thanks
[18:42:33] <Roguish> anyway to 'downgrade'?
[18:42:58] <cradek> I doubt it
[18:43:23] <Roguish> i'm looking at the ubuntu forums, and there are lots of problems...
[18:43:48] <cradek> I think they made a mistake putting that button there...
[18:44:52] <Roguish> i'm probably going to reinstall breezy 'cause emc is all i use this machine for.
[18:45:11] <alex_joni> Roguish: no real way to downgrade
[18:45:17] <cradek> **sigh**
[18:45:59] <Roguish> yeah, i doubt it also. i HATE the upgrade game.
[18:46:04] <Lerneaen_Hydra> I wonder why there isn't downgrade functionality, IIRC there is an apt-get update distro
[18:46:32] <SWPadnos> it's not apt-put downgrade though ;)
[18:46:44] <alex_joni> dist-upgrade it's called.. but should be used with caution
[18:46:55] <Lerneaen_Hydra> alex_joni: can cause broken-ness?
[18:46:57] <SWPadnos> has there not been a successful build of a 2.6.15 based RT kernel?
[18:47:05] <alex_joni> Lerneaen_Hydra: beyond fucked up
[18:47:09] <alex_joni> if I may say that
[18:47:15] <SWPadnos> or is it the packaging that's the problem?
[18:47:15] <Lerneaen_Hydra> giacus compiled one IIRC
[18:47:22] <alex_joni> SWPadnos: jein
[18:47:27] <Lerneaen_Hydra> alex_joni: so really FUBARed then?
[18:47:27] <Roguish> glad it was you
[18:47:29] <SWPadnos> (or is there no problem? ;) )
[18:47:54] <alex_joni> SWPadnos: you certainly know what kind of testing there is needed
[18:50:37] <SWPadnos> I guess it's kinda hard to make one kernel that will work on both distros, and still work for everything else (like udev / dbus / whatever else is needed on dapper, but not necessarily there on breezy)
[18:51:09] <cradek> I don't think that's the goal at all
[18:51:20] <cradek> there is no reason to use a newer breezy kernel
[18:52:07] <SWPadnos> there may be features in dapper that require the newer kernel (though it's not that much newer, so there may not be)
[18:52:33] <cradek> yes there seems to be something wrong with the breezy kernel on dapper
[18:52:51] <cradek> I'm just saying we probably won't upgrade breezy kernels
[18:52:55] <SWPadnos> the other issue is the perception of a downgrade, and the issue that the version number is lower and will probably require shenanigans to install
[18:53:03] <SWPadnos> ok - multiple packages?
[18:53:07] <cradek> we're talking past each other here
[18:53:10] <SWPadnos> heh
[18:53:29] <cradek> yes I think both distribs should have rtai patches to the kernels they came with
[18:53:47] <alex_joni> * alex_joni assumes everyone in here is very deep into emc-developing
[18:53:55] <SWPadnos> ok, so separate packages for each distro, which would render my points moot
[18:54:07] <alex_joni> so they can probably recompile emc2 and even a kernel (given enough nerves & interest)
[18:54:09] <SWPadnos> but makes more ongoing work for the package maintainer
[18:54:13] <cradek> yes they'll be separate
[18:54:20] <cradek> only twice as much :-)
[18:54:21] <alex_joni> SWPadnos: even the maintainers :)
[18:54:21] <Lerneaen_Hydra> wouldn't seperate packages be a hell to maintain?
[18:54:23] <SWPadnos> heh ;)
[18:54:41] <alex_joni> SWPadnos: got a dapper?
[18:54:42] <cradek> I think it'd be worse if one package had to deal with differences in the distribs.
[18:54:51] <SWPadnos> I've got a dapper now
[18:54:51] <Lerneaen_Hydra> with issues/bugs in one but not the other, and differing names to keep track of and so on
[18:55:08] <SWPadnos> cradek, that was what I was pondering before - it didn't look easy
[18:55:16] <alex_joni> SWPadnos: care to test?
[18:55:20] <SWPadnos> sure
[18:55:21] <Lerneaen_Hydra> I guess it's just the least undesireable outcome
[18:55:41] <cradek> SWPadnos: the udev difference we already ran into sure makes me nervous about sharing the packages.
[18:56:03] <SWPadnos> ah - was ther eone? (I was just listing things that I thought could be a problem)
[18:56:16] <cradek> yes
[18:56:19] <SWPadnos> bummer
[18:56:47] <SWPadnos> DBUS may also be an issue, since power management will be disabled
[18:57:10] <SWPadnos> I ran into an interesting issue when reconfiguring the XOrg server in single-user mode
[18:57:13] <alex_joni> SWPadnos: I trust you figure out how to get emc2 for dapper
[18:57:40] <SWPadnos> via package, or self-build?
[18:58:00] <alex_joni> packages
[18:58:12] <Roguish> either
[19:21:17] <alex_joni> jepler: around?
[19:21:44] <jepler> alex_joni: yeah kinda
[19:22:04] <alex_joni> jepler: about to build an axis package
[19:22:34] <jepler> ok
[19:22:36] <alex_joni> or not.. cradek knows something
[19:22:52] <jepler> he's going to tell you they might be compatible from breezy to dapper?
[19:22:58] <alex_joni> probably so
[19:22:59] <jepler> "might"
[19:23:15] <jepler> the odds are not bad, actually
[19:23:19] <alex_joni> hmm.. remembered I use gcc-4.0 not gcc-3.*
[19:23:28] <cradek> oh hmm
[19:23:30] <alex_joni> would that make any difference?
[19:23:35] <jepler> then probably not, there's C++ in there
[19:23:38] <cradek> you better build it then
[19:23:48] <jepler> but you'll notice right away if it doesn't work, because you'll get an ImportError when you run it
[19:23:55] <alex_joni> apt-get source?
[19:24:10] <cradek> oops I should check that change in
[19:24:10] <alex_joni> sounds like the easiest way
[19:24:18] <jepler> you could apt-get -b source .. the same source should build real good
[19:24:41] <alex_joni> * alex_joni mumbles somehting about adding breezy repos
[19:28:05] <alex_joni> cradek: this isn't very fair.. you're making my life way too easy
[19:28:30] <cradek> so it just built with apt-get -b source?
[19:28:44] <alex_joni> running build-dep now
[19:29:12] <alex_joni> -b source now..
[19:29:41] <alex_joni> it builds just fine :D
[19:29:45] <cradek> hmm, you could have also done the emc2 deb that way
[19:29:54] <alex_joni> heh, yeah
[19:30:22] <alex_joni> hmm.. apt-get -b source .. doesn't produce a source package
[19:30:31] <alex_joni> only binary
[19:30:34] <cradek> right
[19:31:02] <alex_joni> hmm.. libglib just updated on dapper
[20:04:20] <lilo> [Global Notice] Hi all. We just had a fairly route loss to one of our largest sponsors. We're back, and so far everything appears to be stable.
[20:10:44] <lilo> [Global Notice] Whoops, that would be "fairly large route looss".... thanks :)
[21:58:53] <Lerneaen_Hydra> Lerneaen_Hydra is now known as Lerneaen_Hydra_a
[21:59:00] <Lerneaen_Hydra_a> Lerneaen_Hydra_a is now known as Hydra_away