#emc-devel | Logs for 2008-11-03

[02:29:18] <jepler> cvs was down for awhile because a gateway machine crashed. service should now be restored.
[02:32:46] <jmkasunich> new dapper non RT VM begins downloading 214 updates....
[02:33:04] <SWPadnos> did you ever get the extra RAM working?
[02:33:14] <jmkasunich> for some value of working
[02:33:24] <jmkasunich> (a value around 3.2G)
[02:33:43] <jmkasunich> server kernel will recognize all 4G, but there is no nvidia driver for that kernel
[02:33:53] <jmkasunich> nv driver has suck
[02:34:15] <jmkasunich> vesa driver is usable, but only has 16 bit memory depth if I want full resolution
[02:34:36] <SWPadnos> you could use Ubuntu 64-bit
[02:34:40] <jmkasunich> (and of course is software GLX, but that seems acceptable for the kind of stuff I do - AXIS and vismach
[02:34:55] <jmkasunich> I have no desire to explore 64bit at the moment
[02:34:58] <SWPadnos> at least with 8.04, even Wine seems to work pretty well on 64-bit
[02:35:10] <jmkasunich> I'm still running 6.06 on this machine
[02:35:23] <SWPadnos> yep, so an upgrade would be a hassle
[02:35:41] <jmkasunich> I'm seriously tempted to jump to 8.04 - the 8.04 server kernal has a linux-restricted-modules package, which should have the nvidia driver
[02:36:12] <SWPadnos> I'm pretty sure the only problems I've had on 64-bit (that I wouldn't have had on 32-bit also) are related to compiling ATI proprietary drivers
[02:45:59] <jmkasunich> hmm, the first time I thought it was a coincidence, but something weird is happening
[02:46:08] <jmkasunich> I have a MyBook USB external hard drive
[02:46:16] <jmkasunich> it is umount'ed, and spun down
[02:46:40] <jmkasunich> when I go into vmware console and create a new vm (click on new, haven't filled in any of the dialogs yet), the drive spins up
[02:46:59] <jmkasunich> doesn't mount, just spins up, later times out and spins down again
[02:47:54] <SWPadnos> it could be doing a partition check
[02:48:53] <jmkasunich> that makes sense (I think)
[02:49:15] <jmkasunich> hmm, I bet dapper doens't actually need 512M of memory once installed
[02:49:30] <SWPadnos> aren't sliders nice? :)
[02:49:42] <jmkasunich> yes
[02:50:01] <jmkasunich> the breezy vm's have 128M, the win95 one has 64M
[02:50:24] <jmkasunich> I just dropped the non-rt dapper (new vm) to 256, I bet I can do the same with the RT one once installed
[02:50:29] <SWPadnos> you can set the VMs to boot into console mode instead of desktop too
[02:50:47] <jmkasunich> I dunno why I thought that I had all the vm's at 512M
[03:42:44] <cradek> neato - the 'at speed' stuff works great
[03:43:18] <cradek> if facing in CSS on the lathe, when it rapids back out to the large radius, it waits for the slowdown before making the next pass
[03:43:35] <cradek> it will also be useful on mills...
[03:44:49] <jepler> so if I hook up a 'timedelay' component on my mill and set the timeout right, I'll get something useful?
[03:45:31] <cradek> to wait for spin-up at spindle on?
[03:45:39] <jepler> yeah
[03:46:04] <cradek> yes that would work since it's single-speed and presumably always takes about the same amount of time.
[03:46:32] <cradek> then it will be able to rapid while spinning up, but it won't feed until the delay has passed
[03:47:33] <cradek> I noticed my actual-spindle-speed display flashes negative agazillion on each threading pass
[03:47:46] <cradek> not a surprised, I suppose - and it must not hurt anything
[03:47:58] <cradek> but a good velocity output from mesa would be nice.
[03:48:02] <jepler> I suppose that makes sense -- it's ddt, not from the driver, right?
[03:48:17] <jepler> if it's from the driver it should be possible to do better
[03:49:15] <cradek> right
[03:49:50] <SWPadnos> does the m5i20 firmware have the timestamp register?
[03:50:01] <SWPadnos> * SWPadnos doesn't think so
[03:50:04] <cradek> I don't think hm4 does
[03:50:09] <cradek> I bet hm2 does
[03:50:12] <SWPadnos> right, unless you guys added it
[03:50:25] <SWPadnos> HM2 has it, but I guess it's not used yet
[03:50:35] <SWPadnos> from PeteW's comments
[03:52:32] <cradek> I cut a whole mess of cast iron on it tonight - was fun
[03:52:42] <jepler> I was talking specifically about velocity estimate over an index reset
[03:53:06] <cradek> I understand - two different concerns
[03:53:29] <SWPadnos> right, I don't think there's any support in the HM4 firmware to make the 5i20 driver any better than ddt
[03:53:40] <jepler> hm2 has two registers for reading the encoder: one is status bits + counter bits, and one is latched counter bits + counter bits. the first one is what is currently being read, I think.
[03:53:44] <SWPadnos> there is in the hostmot2 firmware, but it's unused in the driver so far
[03:53:45] <cradek> I put a dumb velocity output in the driver - it was useless since the spindle encoder is low counts
[03:53:53] <cradek> I run the ddt very slow and it works ok
[03:54:00] <jepler> or maybe there are three, because I just named two and not the one for high-quality velocity estimation
[03:54:23] <SWPadnos> I think it also has a timestamp register so the driver could get accurate counts and time intervals
[03:54:28] <SWPadnos> right :)
[03:55:05] <SWPadnos> index will screw things up at the moment, but it may be possible to just latch the count instead of resetting it
[03:55:27] <SWPadnos> (I think Pete said that was possible with the current firmware, but it's not what the driver does)
[03:56:06] <jepler> 'night
[03:56:08] <cradek> on the cycle when it resets, just output the previous velocity?
[03:56:27] <cradek> seems like it would be fine
[03:56:29] <cradek> goodnight jepler
[03:56:34] <SWPadnos> see you jepler
[03:57:19] <SWPadnos> that would probably work OK for spindles up to 60000 RPM anyway :)
[18:19:29] <seb_kuzminsky> yo
[18:19:38] <jepler> hi seb_kuzminsky
[18:20:01] <seb_kuzminsky> is the firmware packaging i checked in about what you were thinking of?
[18:20:26] <jepler> I didn't look closely, but there is one problem I anticipate: on -sim systems the package probably shouldn't be built or recommended
[18:20:32] <jepler> I think it needs some work in debian/configure to make this possible
[18:21:01] <seb_kuzminsky> right, i see
[18:21:10] <seb_kuzminsky> is there currently a -sim package? i didnt see one
[18:22:50] <jepler> you build the sim package by: debian/configure sim; dpkg-buildpackage -rfakeroot
[18:22:59] <seb_kuzminsky> ah
[18:23:32] <seb_kuzminsky> that debian/configure script confuses me a bit, i'm not used to there being one there
[18:23:40] <jepler> no, it's probably undebianish to do it that way
[18:23:54] <jepler> but unfortunately we don't have debian experts, only me
[18:24:21] <seb_kuzminsky> it seems to be getting the job done, which is not nothing ;-)
[18:24:42] <seb_kuzminsky> i'm no debian expert but i'll take a look at it
[18:25:15] <seb_kuzminsky> would it make sense to compile both emc2 and emc2-sim packages, using different build dirs maybe?
[18:25:34] <alex_joni> seb_kuzminsky: that's only half of the story
[18:25:46] <jepler> well, it's a bit of a mess...
[18:25:49] <alex_joni> we also use configure to get the right kernel for building against the headers
[18:25:56] <jepler> right
[18:26:19] <alex_joni> for example I have 3-4 rtai kernels installed on my box
[18:26:24] <seb_kuzminsky> in case users have multiple rtai kernels installed, for different sandbox RIPs
[18:26:28] <alex_joni> and use debian/configure to point to the right one
[18:26:33] <alex_joni> right
[18:26:37] <SWPadnos> RP isn't sim
[18:26:39] <SWPadnos> RIP
[18:26:39] <jepler> and consider 8.10: there's no realtime kernel, but you can almost certainly build sim
[18:27:52] <alex_joni> another thing I'm not sure about.. how do packages handle multiple distro's?
[18:28:11] <alex_joni> for example building debian (etch or whatever) and ubuntu packages..
[18:28:46] <alex_joni> are there packages out there that do that? or are there branches/forks that do that..
[18:28:47] <jepler> in some cases you can use alternatives in Depends and Build-Depends (the "|") character
[18:28:57] <seb_kuzminsky> alex_joni: i'm not sure how that's done, i think by forking the debian/ dir
[18:29:01] <jepler> Build-Depends: ubuntu-package-name | debian-package-name
[18:29:11] <jepler> lunchtime here
[18:29:17] <jepler> you guys have the debian build system fixed by the time I get back, ok?
[18:29:19] <alex_joni> yeah, but you saw that Build-Depends doesn't work with |
[18:29:27] <alex_joni> only in Depends it works..
[18:29:38] <alex_joni> (remember the lyx stuff we had to move to debian/configure)
[18:29:44] <alex_joni> lyx-qt | lyx
[18:30:12] <seb_kuzminsky> jepler: Yes Sir! ;-)
[18:30:33] <alex_joni> jepler: and rpm's if you have a long lunch
[18:30:58] <seb_kuzminsky> and windows .nsi installers ;-)
[18:31:04] <alex_joni> msi
[18:31:04] <alex_joni> :P
[18:31:10] <seb_kuzminsky> lol
[18:31:27] <alex_joni> it's actually .msi afaik
[18:32:34] <alex_joni> "..structured as OLE Structured Storage Files and commonly known as "MSI files", from their default file extension"
[18:32:47] <seb_kuzminsky> ok i'll do .deb, you do .msi
[18:33:08] <seb_kuzminsky> back to the dayjob for me
[18:33:11] <seb_kuzminsky> later alex
[18:33:23] <alex_joni> later
[19:08:03] <BigJohnT> I've been looking at the document generation using asciidoc to create the xml files and xsltproc to create the xml.fo files then fop to create the html and pdf files
[19:08:16] <BigJohnT> it there anything I need to look out for?
[19:11:42] <alex_joni> tables and pictures mostly
[19:11:48] <alex_joni> text will surely work/look right :)
[19:13:21] <BigJohnT> ok, I'm working on a sample to check the layout and look of both
[19:14:14] <BigJohnT> might take a while :)
[19:14:19] <alex_joni> one thing we could care about is resolution for print/web
[19:14:46] <alex_joni> maybe it works to have 2 different resolutions for the pictures, or the translate process can do that..
[19:15:16] <BigJohnT> yea the stylesheet can take care of different things for html and pdf
[19:17:34] <BigJohnT> I'm about to get a grasp of the whole document chain :O
[19:19:49] <alex_joni> sounds like we have a volunteer to push this :D
[19:20:23] <BigJohnT> * BigJohnT says "who me?"
[19:20:44] <BigJohnT> kinda fun really
[19:21:58] <BigJohnT> well I got a fo with only one error :)
[19:22:41] <alex_joni> fo = following error?
[19:23:39] <BigJohnT> fo formatted output file :)
[19:24:19] <alex_joni> isn't one error usually enough?
[19:24:41] <BigJohnT> it's better than having many!
[20:56:43] <christel> [Global Notice] Hi all, In about five (5) minutes two of our client servers will be going down for maintenance, the downtime window is set to one hour, but in reality it should be significantly less. Affected users ~1200. Thank you for using freenode and have a great day!
[21:11:25] <christel> [Global Notice] Hi all, the upgrades went fine, both servers are now back in rotation -- it went a lot quicker than expected, you'd never guess we run gentoo! Apologies for the inconvenience caused and have a nice day!