#emc-devel | Logs for 2006-09-14

[02:54:44] <jepler> would we give cvs commit access to someone who has contributed an AXIS translation in the past and would like to continue?
[02:57:39] <cradek> I would think yes, if that was the easiest way for the person to contribute
[02:58:24] <cradek> for doing just a translation is it the easiest way?
[02:58:35] <jepler> He may also find that things are not marked in the source
[02:59:02] <jepler> oh well, it's night time here...
[02:59:13] <cradek> let's talk about it tomorrow
[13:51:51] <jepler> Here's what I'm thinking of adding to homing: http://emergent.unpy.net/sandbox
[13:52:01] <jepler> comments?
[14:22:13] <cradek> sounds good except I don't understand the name of the first one, I suggest SHARED_HOME_SWITCH=boolean
[14:23:10] <cradek> DISALLOW_HOME_IF_ON_SWITCH
[14:23:13] <cradek> I dunno
[14:24:13] <cradek> it's hard to be self-documenting and terse both
[14:24:35] <cradek> brb
[14:26:42] <jepler> HOME_SWITCH_SHARED? I think it should have a HOME_ prefix...
[14:45:28] <cradek> I agree
[14:46:09] <cradek> is that the only way it's useful (shared)?
[14:46:25] <cradek> it's all I can think of right now
[15:28:17] <jepler> what do you think of HOME_SEQUENCE?
[15:29:03] <jepler> I think it's good enough for most any cartesian machine
[15:29:37] <jepler> though if you have a regular 3-axis mill with some fixtures you not only want to be sure that Z has been homed before you home X and Y, you want it to be at a set position..
[15:29:38] <cradek> as a way to allow a "home all" button I think it's sufficient
[15:29:50] <cradek> right I was just in the process of typing that.
[15:30:12] <cradek> before I allow axis N to home I need axis 0..N-1 in their home *positions*, not just homed
[15:30:41] <cradek> but that's not a simple gui hack.
[15:31:15] <jepler> there's a flag for "axis is in home position", at least in the motion controller
[15:31:29] <cradek> I didn't know that
[15:32:23] <cradek> this would all have to be in the motion controller. Otherwise I could jog X while Z is homing and screw it up. (it should probably abort, or disallow the jog, or something)
[15:32:55] <jepler> SET_JOINT_AT_HOME_FLAG(joint, 0);
[15:33:00] <cradek> I don't know how complex or foolproof you want to make it
[15:33:05] <jepler> I don't know either
[15:33:48] <jepler> I want a HOME ALL button that will work on a reasonable cartesian machine
[15:33:49] <cradek> the SHARED_HOME_SWITCH is simple, and it's obvious how to do it, and it's very useful to prevent a dangerous and common thing from happing
[15:35:22] <cradek> err HOME_SWITCH_SHARED
[15:36:16] <cradek> I think sequential homing (home all when using combined switches) is a ball/can of wax/mud/worms
[15:36:58] <jepler> maybe so
[15:37:10] <cradek> most of me wants to say "if you want home-all, use separate home inputs (and have a setup that can't cause a crash)"
[15:37:36] <cradek> e.g. you better be above your fixtures when you hit it, and Z better home "up"
[15:38:09] <cradek> but even with switches separate, you can prevent a certain crash by homing Z (up) first
[15:39:06] <cradek> so now I'm back to thinking it's a good idea (assuming you use "homed and in the home position" criteria)
[15:40:01] <jepler> as long as HOME_SWITCH_SHARED is set properly, I don't see that the shared/unshared business matters
[15:40:53] <jepler> AT_HOME is cleared when you jog in free mode, but not if
[15:40:53] <jepler> you do a coordinated move... perhaps that is the intended
[15:40:53] <jepler> behavior.
[15:41:05] <jepler> \todo FIXME - I'm not sure AT_HOME is being reported correctly.
[15:41:20] <cradek> I'd want my sequence to be Z, then X & Y together
[15:41:30] <cradek> that requires separate X,Y switches
[15:41:46] <jepler> yes, but you can do the second-best Z, then X, then Y (or vice versa with X and Y)
[15:42:12] <cradek> so some sequences are just illegal if some switches are shared
[15:42:34] <cradek> in this machine you could share Z and X switches, and have sequence Z, XY
[15:42:39] <jepler> yes, and it would be up to the ini author to make sure it was OK
[15:42:48] <cradek> your flags allow that, right?
[15:42:57] <jepler> yes
[15:43:24] <cradek> ok I think this sounds really good then
[15:43:32] <cradek> assuming AT_HOME works of course
[15:44:22] <cradek> great
[15:45:17] <cradek> I think you'll have to put it all in the motion controller
[15:45:38] <cradek> I mean the home-all smarts
[15:46:29] <jepler> I was hoping I could avoid it
[15:46:50] <cradek> If you have a good spec (I think you do) I bet jmk will help
[15:46:58] <cradek> he's talked about this kind of thing before
[15:47:12] <jepler> I think the first implementation could be in userspace
[15:47:41] <cradek> I don't think you have access to at-home
[15:49:29] <jepler> hmm maybe I don't
[15:49:41] <jepler> I can still compare the position to the inifile HOME_OFFSET
[15:49:57] <jepler> or something janky like that
[15:50:02] <cradek> true
[15:50:30] <jepler> if SEQUENCE=-1 then this joint is not homed by the HOME ALL sequence
[15:52:17] <cradek> on another note, I still want "two-way homing" (search both ways for limit switches, set soft limits inside them, also home
[15:52:46] <cradek> or, at least some way to set soft limits at runtime
[15:53:31] <cradek> think of three buttons like margins on a typewriter: set this as the left soft limit, set this as the right soft limit, temporarily override the soft limits
[15:56:30] <jepler> so before going to HOME_OFFSET you'd go in the same direction until you closed the (other) limit switch, set the soft limit to just short of that value, and go the other way to HOME_OFFSET?
[15:57:36] <jepler> would it make sense to allow axes to be excluded from the HOME ALL sequence (e.g. I don't have any switches on my A axis but I want to home XYZ)?
[15:57:46] <cradek> yes I think that would do it, assuming you home on one limit switch
[15:58:39] <cradek> jepler: without switches it just zeroes, seems like you would want that when you "home all"
[15:58:53] <cradek> but OTOH allowing exclusions doesn't hurt anyone
[15:59:13] <jepler> I imagined that at least one limit would be "fixed" (e.g., the chuck on your lathe doesn't move), and that's the side where the home/limit switch is
[16:00:02] <cradek> yes I think you're right
[16:00:24] <cradek> I wonder if there's a (useful) machine where both limits move WRT home
[16:00:37] <cradek> I mean hard limits here
[16:35:32] <SWP_away> SWP_away is now known as SWPadnos
[16:37:50] <jepler> hi SWPadnos
[16:38:12] <SWPadnos> hi jepler - how are things?
[16:38:28] <jepler> when you're back on the ground I want to talk about getting some access to the webserver to automatically update the html and pdf documentation
[16:39:07] <SWPadnos> ok. I may be able to do that from here
[16:39:15] <jepler> there's no hurry
[16:40:03] <SWPadnos> ok. I may not do that from here ;)
[20:27:38] <alex_joni> odd thing.. had an update on dapper today
[20:27:50] <alex_joni> mostly linux image, and headers for the stock kernel
[20:28:00] <alex_joni> and the updater complained it cannot be verified
[20:29:02] <jepler> huh that is a surprise
[20:29:17] <jepler> maybe it's an evil package put there by a hacker
[20:30:32] <alex_joni> jepler: thought that too
[20:30:41] <alex_joni> did you have an update on the stock kernel lately?
[20:31:21] <alex_joni> linux-image-2.6.15-26-386 is the package name, 2.6.15-26.47 version
[20:31:32] <alex_joni> I found it odd that the changelog doesn't say anything at all
[20:34:31] <jepler> alex_joni: I only have an ubuntu system handly
[20:34:32] <jepler> handy
[20:34:54] <alex_joni> I'll ask in ubuntu-kernel
[20:35:36] <alex_joni> jepler: I think it was because of connectivity loss
[20:35:42] <alex_joni> I did an apt-get update
[20:35:47] <alex_joni> and now the packages are validated
[20:38:26] <skunkworks> alex - this? http://www.electronicsam.com/images/KandT/update.png
[20:39:41] <alex_joni> skunkworks: yes
[20:40:05] <skunkworks> So it is safe? :)
[20:40:11] <alex_joni> although it's still odd.. if you go to 'show details'
[20:40:18] <alex_joni> there's nothing in the changelog
[20:40:22] <alex_joni> it's probably safe
[20:40:34] <alex_joni> I see .47 is the latest version
[20:41:32] <skunkworks> yah - nothing in the change log here also. It has a discription though
[20:41:46] <cradek> sometimes the changelogs show up later
[23:50:17] <jmkasunich> jepler has been busy!
[23:55:13] <jepler> thanks for noticing
[23:56:58] <jmkasunich> the home all stuff looks pretty good
[23:57:30] <jmkasunich> reading back on the other channel I saw you trying to troubleshoot something
[23:57:36] <jmkasunich> did you figure it out>
[23:57:37] <jmkasunich> ?
[23:58:15] <jepler> no -- I was still seeing some kind of random error on startup
[23:58:33] <jepler> it would happen once in awhile
[23:59:19] <jepler> I forget exactly what the message was but you should see it in the scrollback