* SWPadnos kicks CIA-40
I should figure out some reasonable way of making completion work, or get rid of it
let's see if that breaks anything (or anyone)
SWPadnos: halcmd completion is in trunk ?
completion inside halcmd is in 2.2 and TRUNK, completion from the bash command line is in TRUNK only
but it's got an annoying bug that shows up if you try to use it without realtime loaded
how to enable it ?
if you run in place, it should be enabled for you when you source emc-environment
otherwise, take a look at emc-environment, there's a complete command in there (which I don't remember exactly)
(the bash command is called "complete")
who is the cvs admin? cradek?
more like jepler
there's a bunch of now-empty directories that need removal... that takes an admin, right?
this is them:
no, you don't do that
what? list a bunch of stuff to do in CVS and then leave? :)
you rely on 'cvs up -P' to prune directories
that is, remove them in the checkout when there are no files in them with the given tag
they remain on the cvs server forever
can you cvs remove a directory?
that's what I'm telling you: you don't remove directories
if you removed a directory on the server that once contained revisions , you destroy history
of course, you wouldn't delete them on the server
I'm wondering if the "cvs remove" command can tell CVS that a directory is no longer in use
or does it automagically consider "empty" directories to be unused?
$ cvs -H up 2>&1 | grep Prune
ok, so what I'm looking for is more or less automagic
yes, if you put -P in your cvsrc
ie, you do nothing and it works :)
like the wiki suggests
I'm a glutton for punishment - I type it on the command line :)
I finally created a .cvsrc file a few days
but nobody noticed (must have been the inauguration or something distracting them)
looks like quad-core is now affordable :)
toss in this power supply: http://www.newegg.com/Product/Product.aspx?Item=N82E16817703009
underclock most of the time
I think there's only one quad-core at 65W (the 9150e or somthing), all the others are 95+
hmm - not any more. they're all 95W and up
I suppose 125W isn't all that much more than 95W
it's about $3/month more :)
(so it adds up)
there's also the "mips per watt" question, which I don't have the answers to
apparently this errata still affects current-stepping AMD processors: "A DIV Instruction Followed Closely By Other Divide Instructions May Yield Incorrect Results"
"suggested workaround: contact your amd representative for information on a BIOS update"
isn't that a microcode update though (to fix it)?
who knows, it's secret
I wasn't aware that AMDs had reprogrammable microcodes, though I knew intel had it
I can't think of another way that the BIOS could help an errata like that, unless it's something to do with disabling certain chip functions
which might then be re-enabled by the OS anyway ...
[17:14:39] <jepler> http://lwn.net/Articles/291771/
apparently it has
yep, seems that way
heh whos going to put the realtime on the chip then
* archivist ducks
if you get the microcode specs, I'll put some realtime support in the CPU ;)
hmm did they use 290x technology to run intel code
EMC: 03cradek 07TRUNK * 10emc2/debian/changelog: lots more new stuff for 2.3
was it an error in mfg?
I don't think I ever heard back from Mariss but the replacement motherboard fixed the problem for me, so I assume it was a manufacturing/design defect on the REV3 board I had
yes, it was
still is funny that the pwm signal out of mach is 50hz ;)
I scoped the PWM as far as the optoisolator, where I saw little aquare waves going in and noise coming out, so I assume that was the problem :)
50 Hz is fine for a spindle
I should repeat my test of varying PWM frequencies with the G540
I could easily see their filter response on the scope (driving the PWM with a sinewave and changing the freq in halcmd)
hmmm. I wonder if I ever tried PDM mode
cradek: in change log there is typo in motion.spindle-atspeed-in
motion.spindle-at-speed should be
[21:44:56] <micges> http://cvs.linuxcnc.org/cvs/emc2/debian/changelog?rev=1.48
EMC: 03cradek 07TRUNK * 10emc2/debian/changelog: this pin was renamed, thanks micges for noticing