#emc-devel | Logs for 2010-08-20

[00:38:17] <KimK> cradek: Just thought I'd let you know we have run into problems on the BP2 PC due to the ATI Rage 128 (r128) and OpenGL, etc., so we are going to take a run to Microcenter because we think they have a different AGP 4x we can put in there, and see if that helps. If not, we may have to look at compiling our own kernel or something. Gotta go, more news later.
[00:46:47] <ries_> ries_ is now known as ries
[01:03:05] <skunkworks> so - after thinking about it too much - I rewrote the gearshift component. (it was getting to scary and I didn't even get to the lock) and it took me hardly any time and it is simpler. Give me some time and I will get it down to 1 line of code ;)
[01:05:57] <skunkworks> so I settled on - I allow gear shifts on the fly programically. Like if someone changes the s word while the spindle is running. It will not shift gears with the override set to 0 - but if you change to a different gear range while override is 0 - (by changing the s word) it will shift when you move the override above zero.
[01:09:16] <skunkworks> Now the way I arranged the code - spindle lock should be pretty painless to add. When spindle lock goes true - I will just tell the shift section to shift to the right gear - then the lock solenoid will be engaged.
[01:47:54] <cradek> KimK: drop by and pick up a couple old video cards...
[01:48:07] <cradek> too bad we can't fax computer parts or something
[02:06:23] <skunkworks> I could almost lob them there
[03:25:27] <KimK> cradek, skunkworks: OK, back from Microcenter with "new" graphics card, $3.99! (They wanted it out of the store pretty badly, I think.) Anyway, it's in, nVidia proprietary driver (bah!) in, glxgears working, Axis working. And hopefully this will head off future problems too. John said he had gotten help from Chris last year, when 8.04 was first installed on that PC, so it has been an issue.
[03:26:30] <KimK> Thanks again for all your advice and good thoughts.
[03:27:45] <KimK> ATI Rage 128: "I'm b-a-a-a-a-k..." nVidia: "No you're n-o-o-o-o-t..." BLAM!
[03:29:45] <cradek> but the nvidia driver is surely bad news for latency
[03:29:49] <cradek> don't you get the errors?
[03:30:03] <cradek> that driver is a scourge on the earth
[03:30:38] <cradek> $3.99 sounds like the perfect price for a graphics card...
[03:32:47] <KimK> I don't know, didn't check latency, clearly a bad move on my part, I will right now and let you know. (After that wacky "wrong-card" graphics start-up and error boxes, I thought that was the best way out of the mess, but now that it's settled down again, I'd try the generic driver. What, "nv"?
[03:32:59] <cradek> vesa
[03:33:16] <KimK> OK, thanks.
[05:09:14] <ries_> ries_ is now known as ries
[14:04:37] <jepler> hm, I have a half-baked idea that might further improve the software encoder's velocity estimate
[14:05:40] <jepler> for a given true frequency in an ideal system (0 software jitter, 0 signal jitter), the estimate will bounce between two nearest estimates
[14:05:46] <jepler> true+eps1, true-eps2
[14:06:50] <jepler> when you notice you're ping-ponging between two next-door velocities, make the speed estimate the weighted average of those two velocities (weighted according to how often they're measured)
[14:38:12] <jepler> hm, but you need to make the weight a "recent weight" otherwise when the true velocity changes to another velocity that's still between those two values, it's veeerry slow to adapt
[14:56:20] <maximilian_h> Hi
[14:56:30] <maximilian_h> I've got a small new comp
[14:56:34] <maximilian_h> http://pastebin.com/DfQsEzbX
[14:56:59] <maximilian_h> maybe somebody could add it to the collection of comps already present
[14:58:17] <jepler> I think the existing sum2 can do that. It computes out = in0 * gain0 + in1 * gain1 + offset
[14:58:29] <jepler> so just let gain1=1, offset=0
[14:58:35] <maximilian_h> true
[14:58:46] <jepler> er, gain1=1, b is the same as offset
[14:58:52] <maximilian_h> i wasn't thinking about sum2
[14:59:04] <maximilian_h> but your right
[14:59:37] <maximilian_h> but well at least I learned something ;)
[14:59:59] <jepler> I think it's quite valuable to know how to write a comp
[15:00:00] <maximilian_h> Other question
[15:00:49] <maximilian_h> is there an initval in a radiobutton group in pyvcp ?
[15:01:16] <maximilian_h> I would like to make the default not on the first radiobutton
[15:01:28] <maximilian_h> I tried initval, but it was just ignored
[15:02:31] <jepler> looking at the source code (pyvcp_widgets.py) I don't see any way to set the default. It looks like the first one is always selected..
[15:02:34] <jepler> bbl
[15:02:45] <maximilian_h> ok thanks
[15:02:47] <maximilian_h> bye
[16:23:31] <maximilian_h> jepler: Are you there ? If got a patch for pyvcp_widgets.py. Could you apply it ? http://pastebin.com/yKVQn9G1
[16:23:52] <maximilian_h> It's just adding a initval to the radiobutton
[16:25:00] <maximilian_h> value 0 for initval for the first button, 1 for the second button, 2 for the third, etc ad infinitum
[16:25:39] <maximilian_h> anybody else ?
[16:30:15] <maximilian_h> everybody but me seems to be having a beer outside :(
[16:34:23] <seb_kuzminsky> maximilian_h: i dont know anything about pyvcp, but someone will probably look at your patch soon if you mail it to the emc-devel list
[16:35:06] <maximilian_h> seb_kuzminsky: Ok, will try to do that
[16:44:09] <jepler> pastebins are bad for patches anyway (for instance, pastebin.com will give a file with DOS-style line endings when using the DOWNLOAD link, which makes git say 'fatal: corrupt patch')
[16:45:43] <Jymmm> jepler: https://gist.github.com/
[16:47:20] <Jymmm> jepler: See if that works for your needs
[16:47:38] <Jymmm> jepler: has tar, raw, etc
[16:48:05] <maximilian_h> ok, so here it is again http://gist.github.com/540681
[16:49:10] <jepler> thanks
[16:49:51] <CIA-5> EMC: 03jepler 07master * rfb2d3c571132 10/lib/python/pyvcp_widgets.py: pyvcp: radiobutton initial value specification
[16:50:02] <maximilian_h> thanks
[16:50:48] <jepler> you're welcome.
[16:52:34] <maximilian_h> Well, that initval patch makes my life easier :) Bye, have a nice weekend
[16:53:14] <Jymmm> jepler: Have you tried codepad.org?
[16:53:40] <jepler> doesn't ring a bell
[16:53:49] <Jymmm> jepler: you can even test code on it too
[16:55:39] <Jymmm> jepler: http://codepad.org/aNpSknPd
[16:55:57] <jepler> interesting
[16:56:17] <Jymmm> has python too