#emc-devel | Logs for 2010-05-07

[01:05:58] <CIA-2> EMC: 03cradek 07v2.4_branch * r4b410959a4df 10/src/emc/iotask/ioControl.cc: Fix T1M6 / T0M6 / T1M6 not doing the last one, on nonrandom
[01:06:06] <CIA-2> EMC: 03cradek 07master * r72873fe84aff 10/src/emc/motion/ (control.c mot_priv.h motion.c): Export motor offset to HAL
[04:25:53] <CIA-2> EMC: 03seb 07master * r1f75173a3820 10/ (2 files in 2 dirs): Fix hm2 stepgen.enable=0 behavior
[05:11:50] <ries_> ries_ is now known as ries
[12:11:26] <jepler> software stepgen doesn't have its position-fb track its position-cmd while not enabled
[12:13:20] <jepler> emc tracks position-fb and sets position-cmd during machine off, and that avoids gross position changes with the machine off -> machine on transition
[12:16:44] <jepler> so in summary I'm unsure about the portion of this hostmot2 change that changes position-fb and counts behavior
[12:17:14] <jepler> bbl, time to go to the office
[13:56:14] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?action=browse&diff=1&id=Screenshots
[13:56:20] <cradek> touchy got a new look!
[14:03:22] <skunkworks> Neet
[14:04:33] <cradek> hm, in the default (?) theme on lucid, you can't tell which buttons are pressed
[14:04:44] <skunkworks> * skunkworks stayed home to work on the pantry an wait for his laptop to show up
[14:05:24] <cradek> I wish people working on "new looks" would care about usability just the tiniest bit
[14:05:32] <skunkworks> heh
[14:10:01] <cradek> ah I found the answer - keep trying themes until you find one that's sane
[14:11:37] <skunkworks> cradek: did you see this? heard it on public radio and thought you might like the group. http://www.youtube.com/watch?v=Iq4DsqiW2DI
[14:13:17] <cradek> I have never seen someone play a cello while standing up
[14:16:58] <skunkworks> I guess all you need is a longer post.. ;)
[14:17:11] <cradek> he has it tied to himself somehow - you can see at the end
[15:22:30] <micges> I've just compiled emc on 10.04 without any problems
[15:22:53] <JT-Work> nice!
[15:23:08] <micges> system also seems very improved: booting in 7sec, shutting down in 2sec
[15:23:50] <micges> no problems with drivers for brand new hardware
[15:24:25] <micges> even I managed to move closing box to right :D
[15:27:59] <seb_kuzminsky> micges: i love how fast lucid boots
[15:28:37] <seb_kuzminsky> i'm going to add a lucid machine to the buildbot. sim only for now. 64-bit
[15:28:59] <jepler> hi seb_kuzminsky
[15:29:02] <jepler> did you see my criticism earlier?
[15:29:23] <micges> I wanted to try build some rt kernel but didn't get time today, will try on monday
[15:29:25] <jepler> oh, and any idea why one of the buildbots always says it failed?
[15:29:39] <jepler> something about locking apt
[15:30:22] <seb_kuzminsky> logger_dev: bookmark
[15:30:22] <seb_kuzminsky> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2010-05-07.txt
[15:30:37] <jepler> cradek: I see that the class of those problematic widgets is GtkToggleButton so it's perfectly sane to think the two states should be very clearly differentiable.
[15:31:04] <seb_kuzminsky> jepler: there was a crashed apt on one of the buildslaves that prevented an apt build step from running on that vm
[15:31:20] <seb_kuzminsky> i logged in and killed it and the next build on that machine worked
[15:31:40] <seb_kuzminsky> i think it was cron-apt, and i've had problems with that on this virtual cluster before... :-/
[15:32:43] <seb_kuzminsky> jepler: i see the feedback, thanks
[15:34:29] <jepler> seb_kuzminsky: OK
[15:38:56] <jepler> cradek: https://bugs.launchpad.net/ubuntu/+source/light-themes/+bug/537082
[15:40:44] <cradek> tags: added: rhubarb
[15:41:06] <jepler> I wonder what that means
[15:41:42] <jepler> Kenneth Wimer on 2010-05-03
[15:41:42] <jepler> tags: added: rhubarb
[15:41:43] <jepler> removed: gloam
[15:41:55] <jepler> (on a different bug)
[17:21:00] <andypugh> I just noticed Seb leave as I thought of a question I wanted to ask him.
[17:40:51] <cradek> he'll be back. he can't stay away.
[17:40:59] <cradek> (and there's always the dev list)
[19:30:11] <andypugh> Is there a built-in constant for 'eps', ie too close to zero to be sensible?
[20:09:53] <jepler> andypugh: no. what do you want it for?
[20:10:12] <andypugh> divide-by-zero trapping.
[20:10:50] <andypugh> I want to allow sensibly small values, but not too small. I can define it in the file, but I thought there might be a global "standard"
[20:12:05] <jepler> you mean for the "scale"?
[20:12:29] <andypugh> Yes.
[20:14:52] <jepler> For any scale less than 1 there's a value that gives a problematic result (infinity, because it's bigger than the biggest expressible finite number) when you compute value/scale
[20:16:15] <jepler> so you could add an 'if(scale == 0.0)' to test for *exactly* zero but I wouldn't pick any other nonzero cut-off value
[20:16:47] <jepler> or just skip zero protection; I don't see any zero protection in the traditional pwm either
[20:17:02] <andypugh> but _exactly_ zero isn't the only problem.
[20:17:15] <andypugh> Seb said it should be there...
[20:18:50] <andypugh> But you are saying that even with a scale of .5, an input of 2^31+1 is infinite?
[20:19:02] <jepler> what I see as missing in that first patch you submitted is clamping to -1.0 for negative scaled values
[20:19:26] <andypugh> (actually, we only have 9 bits of PWM resolution, so it is worse than that)
[20:19:43] <andypugh> It is there now.
[20:19:47] <jepler> I'm talking about the range of values that can be expressed by a C "double"
[20:19:48] <jepler> >>> 1e300/1e-9
[20:19:48] <jepler> inf
[20:20:12] <jepler> which is bigger than the range of PWM values but still not unlimited
[20:21:05] <andypugh> OK, so check for zero as you can't do anything with NaN, but you can say that -inf <1 ?
[20:21:57] <jepler> -inf < -1 and inf > 1
[20:22:33] <jepler> (I still can't find any protection against scale=0.0 in hostmot2 traditional pwmgen)
[20:22:44] <andypugh> No, it isn't there.
[20:22:58] <jepler> anybody who does that deserves to have to reboot linux :-/
[20:23:01] <andypugh> But Seb said I should have it, and he put it in Stepgen
[20:23:26] <andypugh> And it doesn't seem to cause a serious problem,
[20:33:12] <jepler> it looooks like x/0 will be an infinity with the sign of x, except for 0/0 which is nan.
[20:33:34] <jepler> (under rtai on x86, because all the fpu exceptions such as "infinity" and "divide by zero" are masked)
[20:34:34] <jepler> (fpenv).fxsave.cwd = 0x37f; \
[20:35:38] <andypugh> scale=0 gives 100% PWM regardless.
[20:35:48] <andypugh> Which fits
[20:35:54] <jepler> yes, there is a certain sense to that
[21:58:31] <andypugh> If I have master -> branchA -> branchB and have made a lot of changes to A, what is the process to be able to check branchB back out again?
[21:58:56] <cradek> commit your stuff on A, and then git checkout B
[21:59:18] <andypugh> That's what I assumed, but it is moaning.
[21:59:24] <cradek> what's the moan?
[21:59:33] <andypugh> bear with me
[21:59:41] <andypugh> git gui
[21:59:44] <andypugh> Doh!
[21:59:56] <cradek> bash: git: command not found
[22:01:52] <andypugh> OK, now it didn't complain :-/
[22:03:06] <cradek> are you the one who sent me home-areyousure.diff a long time ago?
[22:03:27] <andypugh> Who me? Not I
[22:03:51] <cradek> hmm
[22:04:24] <micges> I think that was aystarik
[22:04:26] <andypugh> OK, this isn't quite what I wanted as the B-branch doesn't have the changes made to the A-branch
[22:04:51] <andypugh> Including one fairly important file completely gone.
[22:04:58] <cradek> maybe you want to merge A into B, or make a new branch C from the tip of A
[22:05:02] <aystarik> not me. definetly
[22:05:24] <micges> oh ok
[22:05:29] <cradek> forget it, it's already committed, I'm just not paying attention
[22:05:29] <andypugh> I was wanting B to be a branch of A. Perhaps it doesn't work that way.
[22:05:48] <cradek> not sure what you mean by a branch of A
[22:06:18] <andypugh> No, neither am I.
[22:06:20] <cradek> imagebin a picture of gitk --all?
[22:06:21] <cradek> heh
[22:06:44] <cradek> if you don't know what you want, I should stop trying to help :-)
[22:07:22] <andypugh> In terms of git-diffs I wanted master-TPPWM to be the TPPWM stuff only, and TPPPWM-LED to be the LED stuff only.
[22:07:58] <cradek> that sounds linear, not branchy
[22:08:27] <cradek> maybe you have master-TPPWM and master-LED? that's fine, just rebase one onto the other if you want to linearize it
[22:08:42] <andypugh> Yes, but linear is a degenerate form of branchy?
[22:09:12] <cradek> lost me again
[22:09:22] <andypugh> I did have it as I described, but then made a lot of changes to A
[22:09:51] <cradek> what is A?
[22:10:03] <cradek> seriously, I bet a pic of gitk --all would help
[22:10:15] <andypugh> A is TPPWM
[22:10:46] <cradek> I'm too hungry to help with this right now, sorry
[22:11:01] <cradek> I'll help after dinner if you don't get it by then :-)
[23:32:38] <JT-Hardinge> 2.4 pre this file http://pastebin.com/HnsnQrZD skips the tool change to tool 7 for some reason and show tool 6 as the current tool after the T7M6
[23:32:59] <JT-Hardinge> never mind waht a dummy I put M3
[23:33:29] <JT-Hardinge> * JT-Hardinge smacks his forhead on the table again :/
[23:34:20] <andypugh> is X << (32-y) valid or does << only take literals?
[23:35:02] <andypugh> I am finding the error message opaque
[23:35:03] <andypugh> mesa-hostmot2/led.c:115: warning: ISO C90 forbids mixed declarations and code
[23:36:15] <andypugh> (though they are not on the same line)
[23:42:45] <andypugh> Never mind. noob-mistake