#linuxcnc-devel | Logs for 2015-01-15

[00:32:00] <alex_joni> skunkworks: both
[00:36:00] <alex_joni> I mean both good and bad (good there's plenty to do, bad there's no time for other stuff ;)
[12:41:36] <mozmck> When de-referencing a pointer in a structure - like this *(hal->pinthatdoesstuff), is there any real reason to use the () other than readability? If I just use *hal->pinthatdoesstuff I guess it kinda looks to me like it would be de-referencing hal.
[12:49:09] <seb_kuzminsky> using parentheses to show what you want is never wrong, imo
[12:51:43] <mozmck> That's been my general rule too. But *(hal->pinthatdoesstuff) does look funny to me.
[12:53:22] <seb_kuzminsky> looks fine to me ;-)
[13:01:33] <mozmck> I'll get used to it :)
[14:08:56] <skunkworks> heh https://github.com/robEllenberg/linuxcnc-mirror/commit/29d882b0f0fdc63dbd6200a73503d68524cb4a9e
[14:09:35] <skunkworks> I have run through a lot of test programs and no issues...
[14:30:20] <PCW> is that too late for 2.7
[14:42:33] <micges> PCW: I think 2.7 is waiting for those fixes
[15:25:49] <mozmck> hmm, steinkuehler made a gantry component that looks interesting for a possible quick fix: https://www.youtube.com/watch?v=7XhIoDV8Hp8
[15:56:46] <mozmck> cradek: does haberler's list of problems with jaX look right to you? interesting that he posted this just after we started looking at it :) https://github.com/machinekit/machinekit/issues/438
[15:58:20] <cradek> he's right about wheel axis jogs not being done
[15:59:02] <cradek> I thought UI jogging worked, maybe he's talking about their own UIs, hard to say
[15:59:29] <cradek> kins, no idea, inverse-only has been moderately broken forever, probably not a ja-specific problem
[15:59:30] <mozmck> what is wheel joggin?
[15:59:45] <cradek> you know, the realtime jogwheel thing
[15:59:53] <cradek> it jogs joints
[16:00:00] <mozmck> mpg?
[16:00:00] <cradek> ideally it would be able to jog axes too
[16:00:04] <cradek> yeah that
[16:00:08] <mozmck> i see.
[16:00:19] <cradek> but that's new functionality that would be nice, and not a reason to not merge, IMO
[16:00:31] <mozmck> what about jogging more than one axis at a time?
[16:00:32] <cradek> joint-only wheel jogging would be no worse than today
[16:00:43] <mozmck> Ok.
[16:00:44] <cradek> don't know, let me know what you find
[16:01:06] <cradek> I also think there's something wrong with configs having nonconsecutive axes
[16:01:24] <mozmck> does that work fine in master? he indicates it is a lack in NML
[16:01:37] <cradek> diagonal joint jogs? certainly it works in master
[16:01:58] <cradek> diagonal axis jogs I don't know
[16:02:24] <mozmck> I'm guessing like jogging XY at the same time...
[16:03:07] <cradek> do you mean XY (nontrivkins) or do you mean more than one joint (like holding down two arrows in trivkins)
[16:03:13] <cradek> I know the latter works
[16:03:21] <mozmck> Did you see the gantry comp I linked to? https://groups.google.com/forum/#!searchin/machinekit/gantry/machinekit/To6ivNZpEHU/eamHoiO_868J
[16:04:00] <mozmck> I'm guessing he means two axes instead of 2 joints.
[16:05:31] <cradek> I haven't looked at that comp
[16:05:48] <cradek> I am not very interested in a solution that doesn't support all the homing types including index
[16:05:57] <cradek> but it might be a fine workaround until we have that
[16:06:13] <mozmck> That's what I was thinking/wondering.
[16:08:16] <mozmck> Oh yeah, I kind of remember this: "When a joint home switch trips, the commanded
[16:08:16] <mozmck> velocity will drop immediately from HOME_SEARCH_VEL to zero, with no limit on
[16:08:16] <mozmck> accleration.
[16:08:19] <mozmck> "
[16:09:08] <cradek> hm, yeah I think the smarts really needs to be inside motion to work well
[16:09:23] <mozmck> sounds like.
[16:09:33] <cradek> you can get away with a lot on really light stepper machines
[16:10:22] <mozmck> true. most of the stuff we deal with is medium weight/size though.
[16:48:16] <skunkworks> zlog