Back
[02:15:29] <nojepler> the leap second "should" be OK, because CLOCK_MONOTONIC doesn't care about them, and we use CLOCK_MONOTONIC for timing in uspace rtapi_app.
https://unix4lyfe.org/leap-second/
[02:17:32] <nojepler> 96boards hikey seems still not to be shipping
[02:18:47] <nojepler> (they had said in May that orders through April would be fulfilled by the first week of June)
[08:38:59] <jthornton> is GNU GPL v3 compatible with us?
[08:39:12] <jthornton> https://code.google.com/p/libarea/
[08:44:16] <jthornton> never mind it looks over my pay scale
[09:27:23] <andypugh> seb_kuzminsky: Do you want the carousel.comp? If so, which branch?
[10:02:04] <cradek> andypugh: would you commit the whole sample config with the vismach model? it's very neat and useful.
[10:08:02] <andypugh> OK, I can do that.
[10:08:33] <andypugh> But the question of which branch remains :-)
[10:09:16] <andypugh> The Vismach model as-is depends on orient having the is-oriented pin and that has only been added in Master
[10:11:26] <andypugh> I could try to make it work wth “near” but the timing between a new orient-angle from motion, a new target from orient, and then motion not seing the “near” component tell it that is has alrady finished is a puzzle I haven’t solved (hence the addition of a pin to orient)
[10:34:26] <KGB-linuxcnc> 03andy pugh 05master a91a779 06linuxcnc 10src/hal/components/orient.comp Add a default value to the is-oriented tolerance. * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a91a779
[11:45:53] <Gurwinder_> cmorley: Hi
[16:08:22] <skunksleep> Dead
[16:22:44] <Tom_itx> deadskunk?
[16:22:46] <andypugh> Are you? How unfortunate
[16:27:44] <skunksleep> https://youtu.be/4WyqzdDPHYA
[16:28:05] <skunksleep> Not much activity today
[16:28:33] <skunksleep> Looks like he has some kins running... Cool
[16:35:01] <andypugh> Cool. if only we could limit joint speeds properly
[16:38:42] <skunksleep> Yes. There was some discussion on the machine kit side about this... It sounds hard:)
[16:42:12] <skunksleep> Seems if you could do the calcs in kins to set a maxvel or adaptive feed.
[16:42:51] <micges> it's not hard, rather tricky
[16:43:12] <andypugh> You need to store the previous kins results to find joint velocities at the current position.
[16:43:14] <skunksleep> Based on joint limits. But that doesn't help acceleration
[16:44:10] <andypugh> capping velocity would be a good start though.
[16:45:50] <skunksleep> Smop
[16:51:14] <andypugh> Does three callls to kins with a small delta-t give accel?
[16:53:31] <skunksleep> But what are you going to do with it? I don't know how you would 'scale' acceleration..
[16:54:26] <skunksleep> Velocity I can see.. But not acceleration.
[17:02:01] <skunksleep> Or would the process of scaling velocity have the same effect on acceleration? So you would just check acc and vel and then just scale velocity down until either one is satisfied.
[17:04:10] <skunksleep> That almost seems plausible
[17:05:02] <andypugh> Accel tells you how fast you can back-off the velocity
[17:07:38] <skunksleep> Maxvel and adaptive feed are pretty kins. Hmmm
[17:09:42] <skunksleep> Pre...
[17:41:43] <Roguish> gotta be really careful with angular accel. there's some squares in it and it can tip velocities can get big pretty fast.