#emc-devel | Logs for 2008-11-23

[00:01:11] <tissf> better for you to send the files with your big bike :)
[00:01:22] <BigJohnT> cool
[00:01:45] <BigJohnT> getting too cold to ride even with a built in heater
[00:02:33] <alex_joni> hi guys
[00:02:48] <alex_joni> * alex_joni catches up on lots of email
[00:02:57] <tissf> put your wife on the handlebars and climbed into the sidecar
[00:03:05] <BigJohnT> hi alex
[00:03:14] <alex_joni> hey BigJohnT
[00:03:15] <tissf> hello Alex
[00:03:16] <BigJohnT> she rides on the back LOL
[00:03:19] <alex_joni> salut tissf, ca va?
[00:03:40] <tissf> fine thanks
[00:04:04] <alex_joni> * alex_joni just got home
[00:04:09] <alex_joni> been a long but excitign day
[00:04:21] <tissf> ahah?
[00:06:19] <tissf> alex_joni: another robot?
[00:06:52] <alex_joni> tissf: no.. just a trip to belgrade today
[00:06:56] <alex_joni> having some fun for a change
[00:07:24] <CIA-38> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (Master_Integrator.lyx Submakefile): add file for AX5214H and opto22 driver
[00:07:34] <CIA-38> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/drivers.lyx: add file for AX5214H and opto22 driver
[00:08:04] <BigJohnT> on a skiing trip in the Alps?
[00:08:54] <alex_joni> BigJohnT: missed by a bit ;)
[00:09:07] <alex_joni> no alps around belgrade .. just sightseeing
[00:09:20] <alex_joni> it's about 3h away from here, and I've never been there to visit it properly
[00:09:42] <BigJohnT> belgrade?
[00:09:46] <tissf> :) it was snowing today in Belgrade?
[00:13:17] <tissf> BigJohnT: will there still a lot of change in the doc of the ladder?
[00:13:44] <BigJohnT> I should be done unless someone finds a typo or spello
[00:14:26] <alex_joni> tissf: it should have been snowing.. I only got 2-3 snowflakes though, so it was ok
[00:14:29] <alex_joni> pretty cold though..
[00:14:49] <alex_joni> BigJohnT: .......
[00:14:55] <alex_joni> http://en.wikipedia.org/wiki/Belgrade
[00:15:38] <tissf> alex_joni: same here
[00:16:35] <tissf> BigJohnT: this is the last file to be translated into French
[00:17:02] <alex_joni> well.. I'm off to bed (20h days are too long ;)
[00:17:09] <alex_joni> good night all
[00:17:17] <BigJohnT> good night alex
[00:17:27] <tissf> alex_joni: good night Alex
[00:18:43] <tissf> John you seen my videos on the wiki?
[00:18:56] <BigJohnT> no, which ones are they?
[00:19:51] <tissf> http://www.youtube.com/watch?v=KGMF61LHHww
[00:20:04] <tissf> http://www.youtube.com/watch?v=AyB3HJlDNCo
[00:20:41] <tissf> with a probe that I made
[00:21:17] <BigJohnT> it will take me a while to down load :(
[00:21:57] <tissf> you have an RTC?
[00:22:01] <BigJohnT> is it like a Renashaw probe?
[00:22:36] <tissf> yes same principle
[00:22:40] <BigJohnT> rtc?
[00:22:55] <BigJohnT> cool
[00:22:58] <tissf> no DSL ?
[00:23:40] <BigJohnT> what is DSL? and RTC?
[00:24:02] <tissf> you're really in the woods?
[00:24:05] <tissf> :)
[00:24:52] <BigJohnT> very much in the woods, I live on a dead end dirt road off of a dirt road down in a holler
[00:25:11] <BigJohnT> and you can't see the dirt road from my house...
[00:27:37] <tissf> you live in a ranch?
[00:28:41] <BigJohnT> I have 9 acres of woods not really that big...
[00:29:39] <tissf> a small ranch with some horsepower
[00:29:56] <tissf> :)
[00:31:04] <BigJohnT> dog power here, my neighbor has some horses
[00:32:45] <tissf> OK, John, bed time here
[00:32:46] <BigJohnT> dinner time here talk to you later
[00:32:55] <tissf> :))
[00:33:14] <BigJohnT> good night Francis
[00:33:31] <tissf> good night John
[04:33:31] <jmkasunich_> hmm, I didn't know halcmd show pin accepted a type specifier
[04:33:42] <SWPadnos> it does?
[04:33:44] <jmkasunich_> show pin -t bit
[04:33:47] <SWPadnos> cool
[04:34:01] <jmkasunich_> it isn't documented, but it works
[04:34:02] <SWPadnos> * SWPadnos didn't know either :)
[04:34:15] <SWPadnos> how about param?
[04:34:36] <jmkasunich_> looks like the same code is called for pin, sig, and param
[04:34:43] <SWPadnos> cool
[04:35:16] <jmkasunich_> I suppose show alias should do the same?
[04:35:39] <SWPadnos> well, if it's easy yes
[04:35:43] <jmkasunich_> I also suppose that there should be "show alias pin" and "show alias param"
[04:36:03] <SWPadnos> yes
[04:36:32] <SWPadnos> there could also be "all", which might be the default if no entity type is specified
[04:36:48] <SWPadnos> though that's a halcmd thing, not a library thing
[04:37:14] <jmkasunich_> yes, I'm working on the halcmd side
[04:37:20] <SWPadnos> oh, ok
[04:37:24] <jmkasunich_> can't test the lib part otherwise
[04:37:32] <SWPadnos> I should see if I have anything worth committing there
[04:37:49] <SWPadnos> I think I only had added a function stub
[04:38:02] <jmkasunich_> for the actual alias command?
[04:38:13] <SWPadnos> yes, but it wouldn't have worked
[04:38:30] <SWPadnos> oh, maybe it would have
[04:39:10] <jmkasunich_> damn
[04:39:16] <jmkasunich_> looks like my RT computer died
[04:39:31] <jmkasunich_> no lights, no response to power switch
[04:40:19] <SWPadnos> it was of type ugh
[04:40:21] <SWPadnos> err
[04:40:23] <SWPadnos> ugh
[04:40:39] <SWPadnos> unplug it and try hitting the switch again
[04:40:47] <SWPadnos> (well, after plugging it back in)
[04:41:01] <jmkasunich_> I've cycled the genuine hardware switch on the back of the PS
[04:41:26] <SWPadnos> if it seems to come on for a half second then die (you may only notice this from the fans trying to spin), then the PS is crowbaring
[04:42:01] <SWPadnos> and you probably have a shorted something-or-other somewhere
[04:42:34] <jmkasunich_> absolutely nothing
[04:42:59] <SWPadnos> hmm
[04:43:03] <jmkasunich_> I'll try again when the furnace stops running, so I can hear any tiny fan noises, but it seems totally dead
[04:43:44] <SWPadnos> well, I'll say this - I tried a different power supply in my computer, and its crowbar wasn't as good or something, because that's when the smoke came out of a CPU power supply transistor
[04:43:56] <SWPadnos> it glowed nicely first though
[04:44:22] <jmkasunich_> ouch
[04:44:33] <jmkasunich_> oh well, I'll deal with that tomorrow
[04:44:49] <jmkasunich_> I can do the alias stuff on this (sim) machine
[04:47:51] <jmkasunich_> weird, it just started up
[04:48:34] <jmkasunich_> ick, 18 orphan inodes
[04:48:39] <SWPadnos> bummer
[04:48:53] <SWPadnos> did it start on its own or did you hit the button again?
[04:48:58] <jmkasunich_> I hit the button
[04:49:04] <SWPadnos> phew
[04:49:07] <jmkasunich_> it started on the 2nd or 3rd attempt
[04:49:36] <SWPadnos> I've seen machines not want to power on until you hold the power button for 4 or 5 seconds
[04:49:44] <SWPadnos> like the delayed power off function
[04:49:50] <SWPadnos> but for power on
[04:49:53] <jmkasunich_> that might have been the case here
[04:50:10] <jmkasunich_> I pushed the button several times the normal way, then held it for a couple seconds=
[04:50:18] <jmkasunich_> the fan started on button release
[04:50:31] <SWPadnos> hmm
[04:52:13] <SWPadnos> ah ok - it wouldn't work
[04:52:29] <SWPadnos> alias needs 3 arguments, and there's no A_THREE type yet
[04:52:47] <jmkasunich_> thats the part that gives me fits
[04:52:59] <jmkasunich_> I'm gonna work on show alias instead
[04:53:18] <SWPadnos> ok
[04:53:38] <SWPadnos> I din't think I added that to the completion code - I'll do that now
[04:53:57] <jmkasunich_> show alias? I just did that
[04:54:04] <SWPadnos> no, the completion for it
[04:54:12] <jmkasunich_> static const char *show_table[] = {
[04:54:13] <jmkasunich_> "all", "comp", "pin", "sig", "param", "funct", "thread", "alias",
[04:54:17] <SWPadnos> ah, yep
[04:54:26] <SWPadnos> also save (for future use)
[04:54:45] <jmkasunich_> let me do those, so our commits don't conflict
[04:54:56] <SWPadnos> and list
[04:54:58] <SWPadnos> ok
[04:56:02] <SWPadnos> have you modified halcmd.c?
[04:58:52] <jmkasunich_> I don't have any uncommitted changes to it
[04:59:26] <SWPadnos> ok
[05:00:26] <SWPadnos> ok, I'm in halcmd_commands.c and halcmd.c
[05:00:43] <jmkasunich_> I wonder what show alias should print? I was thinking just alias and original name
[05:00:55] <SWPadnos> yep
[05:00:58] <jmkasunich_> do we want type, owner, value, etc, same as show pin? I'd rather not
[05:01:06] <SWPadnos> I think that should just be about the names, not the other info
[05:01:10] <jmkasunich_> good
[05:01:14] <SWPadnos> since show pin will show that
[05:01:23] <jmkasunich_> now, should the pattern apply to the original name, or the alias?
[05:01:36] <SWPadnos> for show?
[05:01:36] <jmkasunich_> or both (if either matches, print it)
[05:01:39] <jmkasunich_> yes
[05:01:58] <SWPadnos> hmmm
[05:02:09] <SWPadnos> I wouldn't want to see things twice
[05:02:16] <jmkasunich_> you won't
[05:02:24] <jmkasunich_> it will traverse the list once
[05:02:49] <jmkasunich_> but when deciding "do I print this node", it will compare the pattern against either original name, alias, or both
[05:02:55] <SWPadnos> bith
[05:02:56] <SWPadnos> both
[05:03:31] <jmkasunich_> so you can ask "what is the original name of foo", as well as "what is foo aliased to"
[05:03:46] <SWPadnos> hmmm
[05:03:49] <SWPadnos> well, I don't know
[05:04:20] <SWPadnos> it may be confusing if you "show pin hm2*" and you see things that don't print when you do "show pin"
[05:04:34] <jmkasunich_> no no no
[05:04:40] <jmkasunich_> I'm talking about show alias
[05:04:42] <SWPadnos> since in one case you'll see on ly the aliased name, and in the other the original name could match
[05:04:50] <SWPadnos> oh, right :)
[05:05:26] <jmkasunich_> I'm doing this in bite sized pieces - for now, I just need show alias
[05:05:30] <jmkasunich_> then I create some aliases
[05:05:36] <SWPadnos> right!
[05:05:44] <jmkasunich_> then I modify show pin to show the alias names instead of the orig names
[05:05:58] <SWPadnos> ok, let me commit the (very little) skeleton I have for the alias command
[05:06:09] <jmkasunich_> ok
[05:09:35] <SWPadnos> hmmm - one sec. it's just a prototype, but I suppose I should make sure it's getting called as expected :)
[05:13:33] <CIA-38> EMC: 03swpadnos 07TRUNK * 10emc2/src/hal/utils/ (halcmd.c halcmd_commands.c halcmd_commands.h): start of adding alias command - alias now gets called with the correct number of arguments
[05:13:34] <SWPadnos> man, when did the default editor for commit messages become vi
[05:13:36] <SWPadnos> ?
[05:13:46] <jmkasunich_> always been I think
[05:13:52] <jmkasunich_> I always use commit -m
[05:14:00] <SWPadnos> no, I used to have nano or something usable
[05:14:14] <jmkasunich_> EDITOR maybe?
[05:14:20] <fenn> VISUAL
[05:14:23] <SWPadnos> hmmm, could be
[05:14:32] <jmkasunich_> ISTR that my dapper systems invoked nano, and hardy invoked vi
[05:14:39] <jmkasunich_> fenn: VISUAL is a gui editor
[05:14:45] <fenn> oh.
[05:14:46] <jmkasunich_> EDITOR gets you a text mode one
[05:15:00] <SWPadnos> neither is set. maybe I should do that ;)
[05:17:17] <jmkasunich_> jmkasunich_ is now known as jmkasunich
[05:20:14] <jmkasunich> your changes merged without conflicts
[05:23:16] <SWPadnos> cool
[05:23:38] <SWPadnos> all that happens is the function gets called with type, name, alias
[05:23:54] <jmkasunich> weird, gcc thinks my function ends about in the middle of the code
[05:24:00] <SWPadnos> with all the error checking that comes before that, like too many or too few parameters
[05:24:04] <jmkasunich> the editor says my braces are matched.....
[05:24:06] <SWPadnos> hmm
[05:24:23] <SWPadnos> any nested comments?
[05:24:44] <jmkasunich> no
[05:25:26] <jmkasunich> duh, mismatched ()
[05:25:34] <SWPadnos> heh
[05:26:49] <jmkasunich> if I'm not displaying the type or any other info except name and alias, I don't think I want the -t type filter
[05:27:39] <SWPadnos> nope
[05:28:09] <jmkasunich> when I'm doing show all, should I print the pin aliases immediately after the pins, and param aliases right after params
[05:28:14] <jmkasunich> or should all aliases be at the end?
[05:28:32] <SWPadnos> I think they should be separate sections
[05:28:51] <SWPadnos> at the end is fine, since they're not "functional"
[05:41:21] <jmkasunich> so far so good - I created an alias, show alias shows it, and net can link a signal to it
[05:42:08] <SWPadnos> nice
[05:43:11] <jmkasunich> hmm, the completion for things like net is gonna need changed
[05:43:19] <jmkasunich> it should complete aliases as well as pin names
[05:43:41] <SWPadnos> yep - I can take care of that
[05:44:52] <jmkasunich> halcmd: show alias
[05:44:52] <jmkasunich> Pin Aliases:
[05:44:52] <jmkasunich> Alias Original Name
[05:44:52] <jmkasunich> enable2 pid.2.enable
[05:44:52] <jmkasunich> maxerror2 pid.2.maxerror
[05:45:12] <SWPadnos> nice
[05:45:26] <jmkasunich> now to alter show pin
[05:45:53] <SWPadnos> can you commit that - I can fiddle with completion
[05:46:16] <jmkasunich> lemme check the diff...
[05:48:14] <CIA-38> EMC: 03jmkasunich 07TRUNK * 10emc2/src/hal/hal_lib.c: aliases - work in progress, commit for collaboration
[05:48:21] <CIA-38> EMC: 03jmkasunich 07TRUNK * 10emc2/src/hal/utils/ (halcmd_commands.c halcmd_completion.c): aliases - work in progress, commit for collaboration
[05:48:58] <SWPadnos> thankw
[05:49:00] <SWPadnos> s
[05:54:56] <jmkasunich> ick
[05:55:01] <jmkasunich> the merge sort is messy
[05:55:21] <SWPadnos> yeah, I got a little lost in the details when thinking about it :)
[06:06:04] <CIA-38> EMC: 03seb 07TRUNK * 10emc2/configs/hostmot2/hm2.hal: sample config now uses the new GPIO pin names
[06:06:43] <CIA-38> EMC: 03seb 07TRUNK * 10emc2/docs/man/man9/hostmot2.9: describe encoder.reset better
[06:07:39] <CIA-38> EMC: 03seb 07TRUNK * 10emc2/src/hal/drivers/mesa-hostmot2/ (TODO encoder.c hostmot2.h):
[06:07:39] <CIA-38> EMC: This adds basic reciprocal-time encoder velocity computation.
[06:07:39] <CIA-38> EMC: Still freaks out on index for now...
[06:07:44] <garage_seb> cradek: it's a start
[06:31:07] <jmkasunich> vile:
[06:31:08] <jmkasunich> curr_alias = hal_data->pin_alias_list_ptr ? SHMPTR(hal_data->pin_alias_list_ptr) : NULL;
[06:31:33] <jmkasunich> curr_alias = curr_alias->next_ptr ? SHMPTR(curr_alias->next_ptr) : NULL;
[06:36:04] <SWPadnos> heh. I'm trying to figure out how to run halcmd again after a segfault (in sim)
[06:36:51] <jmkasunich> -R ?
[06:37:20] <SWPadnos> oh hey, that worked
[06:37:22] <SWPadnos> thanks
[06:44:30] <jmkasunich> show pin seems to work, aliased pins are displayed using their alias, in the right order
[06:47:08] <SWPadnos> great
[06:47:18] <jmkasunich> I'm not sure what comes next
[06:47:23] <jmkasunich> I'm sure something is missing
[06:47:26] <SWPadnos> I guess I'm writing the same code right now, for completion to work
[06:47:40] <SWPadnos> oh, hmm
[06:47:45] <SWPadnos> there's no way to unlias
[06:47:55] <jmkasunich> that can be simpler, can't it? completion doesn't need to be sorted
[06:48:03] <SWPadnos> alias requires exactly 3 arguments
[06:48:12] <SWPadnos> completion should be sorted I think
[06:48:18] <SWPadnos> not required, but better
[06:48:32] <jmkasunich> for people who hit double-tab to see the list?
[06:48:54] <SWPadnos> yep
[06:49:13] <jmkasunich> I wonder - does the code that actually implements completion do a sort?
[06:49:32] <SWPadnos> no
[06:49:39] <jmkasunich> or does it brute-force filter the list whenever you add another character
[06:49:42] <SWPadnos> well, the pin generator doesn't anyway
[06:50:01] <jmkasunich> when I say "code that implements", I mean the library stuff
[06:50:06] <SWPadnos> I don't know
[06:50:11] <SWPadnos> I don't think so
[06:50:27] <SWPadnos> there's a way to find out though :)
[06:50:33] <jmkasunich> why don't I commit this, so you can see how I did it
[06:51:26] <SWPadnos> ok, thanks
[06:51:39] <SWPadnos> then again, I'm pretty sure I'm going to bed soon :)
[06:51:48] <CIA-38> EMC: 03jmkasunich 07TRUNK * 10emc2/src/hal/utils/halcmd_commands.c: aliases - show alias and show pin now working, commit for collaboration
[06:52:20] <jmkasunich> it sucks that we wind up doing that merge sort everywhere we need a pin list
[06:52:29] <SWPadnos> yeah
[06:52:32] <jmkasunich> halcmd show, halcmd completion, halmeter, halscope
[06:52:52] <jmkasunich> maybe I should actually re-order the pin list in hal_lib - solve the problem in one place
[06:53:12] <SWPadnos> well, it depends though
[06:53:13] <CIA-38> EMC: 03seb 07TRUNK * 10emc2/docs/man/man9/hostmot2.9: Add support for three of the four hm2 step_types: step/dir, up/down, and quadrature
[06:53:13] <CIA-38> EMC: 03seb 07TRUNK * 10emc2/src/hal/drivers/mesa-hostmot2/ (TODO hostmot2.h stepgen.c): Add support for three of the four hm2 step_types: step/dir, up/down, and quadrature
[06:53:27] <jmkasunich> depends on what?
[06:53:30] <SWPadnos> I guess all those consumers of lists of pins only want the alias or the name, not both
[06:53:40] <jmkasunich> right
[06:53:51] <SWPadnos> whether the alias should replace the name for the most part, or whether it's an additional name (an alias :) )
[06:54:07] <jmkasunich> well, we've been assuming that it replaces the name
[06:54:24] <jmkasunich> in fact, I just realized a way to make this a crapload simpler
[06:54:31] <SWPadnos> ok, so if completion should match one or the other, then that makes it easier
[06:54:34] <SWPadnos> oh, good! :)
[06:54:49] <jmkasunich> the alias object should not store the alias name, it should store the original name
[06:55:09] <jmkasunich> the alias name should go in the pin object itself, and then the object gets re-ordered
[06:55:18] <jmkasunich> that makes aliases invisible to other code
[06:55:34] <SWPadnos> ok, then add an unalias command
[06:55:47] <jmkasunich> that was gonna be needed anyway
[06:55:47] <SWPadnos> that knows how to look for aliases or old names
[06:55:59] <SWPadnos> yeah, I think so
[06:56:08] <jmkasunich> the hal_lib API already can do unalias, just pass null as the name
[06:56:38] <SWPadnos> right - it's just a little harder (not too much) to make the alias command accept zero or one aliases
[06:56:48] <SWPadnos> easier to have a fixed number of arguments
[06:56:57] <jmkasunich> in halcmd, there should be a distinct unalias command
[06:57:06] <SWPadnos> I agree
[06:57:17] <jmkasunich> and it should call the API function with NULL
[06:57:30] <jmkasunich> that is completely independent of the internal implementation of aliases
[06:58:00] <jmkasunich> dang, storing the old name makes a ton of sense
[06:58:14] <jmkasunich> I don't even need to link those alias records into lists
[06:58:48] <jmkasunich> show alias just runs the pin list, printing only entries that have orig-names linked to them
[06:59:07] <jmkasunich> the list is already sorted by active-name
[06:59:15] <jmkasunich> (or will be, once I do the hal_lib part)
[06:59:29] <SWPadnos> you still need to keep track of used alias space though
[06:59:51] <jmkasunich> you mean for the status command?
[07:00:03] <SWPadnos> no, to keep track of memory
[07:00:12] <SWPadnos> status benefits also
[07:00:28] <jmkasunich> memory management doesn't need the alias lists
[07:00:40] <SWPadnos> oh, ok then
[07:00:42] <jmkasunich> alias objects are freed when the associated pin is freed
[07:01:12] <jmkasunich> the only reason alias objects were linked into a list (in addition to being linked to their associated pin) was for "show alias"
[07:03:20] <jmkasunich> I'm gonna start over (not quite) tomorrow
[07:03:31] <SWPadnos> ok. me too I think
[07:03:37] <SWPadnos> well, except for the starting over part ;)
[07:03:41] <jmkasunich> don't do anything for the completion lists - I think they will be fine with no changes
[07:03:47] <SWPadnos> I'll just change all this back before then
[07:03:59] <jmkasunich> in fact, I bet 80% of what I did today will wind up being reverted
[07:04:09] <SWPadnos> rm halcmd_completion.c && cvs up ;)
[07:04:35] <jmkasunich> goodnight
[07:04:37] <SWPadnos> night
[07:44:15] <CIA-38> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/ (8 files):
[07:44:15] <CIA-38> EMC: Created new variables IW and QW to represent S32 in and out pins instead of
[07:44:15] <CIA-38> EMC: mapping them to W variables. Defaults to ten of each. Aligns us with mainsteam
[07:44:15] <CIA-38> EMC: Classicladder. ladder programs will need to be updated, HAL files will be fine
[08:41:51] <CIA-38> EMC: 03cmorley 07TRUNK * 10emc2/docs/man/man9/classicladder.9: Update classicladder man page a little
[10:08:46] <CIA-38> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/spy_vars_gtk.c: Fix integer variable window to display labels again after adding the IW and QW variables
[15:49:35] <CIA-38> EMC: 03flo-h 07TRUNK * 10emc2/src/po/de_rs274_err.po: updated
[16:07:45] <alex_joni> hmm.. si email acting up? didn't see flo-h's commit email
[16:12:40] <Guest435> Guest435 is now known as skunkworks
[16:14:35] <SWPadnos> alex_joni, email is acting normal, we may see the commit message in a few days :)
[16:19:22] <SWPadnos> oh, there it is now
[16:19:46] <SWPadnos> looks like it hit on the 30-minute retry cycle
[16:27:00] <CIA-38> EMC: 03seb 07TRUNK * 10emc2/docs/man/man9/hostmot2.9: note stepgen quadrature mode behavior
[17:08:18] <alex_joni> why doesn't this work with emc2? *whiine* http://www.ar.com.au/~softmark/page44.html
[17:09:43] <skunkworks> well - why doesn't it? Huh?
[17:10:00] <alex_joni> I've never seen a more stupid webpage
[17:11:15] <SWPadnos> at least they're smart enough to use an AVR or something
[17:11:19] <SWPadnos> or a 20-pin PIC
[17:13:22] <SWPadnos> oh, maybe it's a MC68HC908JB8JPE
[17:23:38] <alex_joni> wee.. motorolla :)
[17:23:49] <alex_joni> s/ll/l/
[17:30:25] <SWPadnos> ahem, that's Freescale now :)
[17:32:53] <alex_joni> yeah, well.. whatever :P
[19:03:29] <Guest887> hi all
[19:46:39] <alex_joni> hmm.. do theemc2 tests fail for you tooo?
[19:48:33] <cradek> Runtest: 25 tests run, 24 successful, 0 failed + 1 expected
[19:49:08] <alex_joni> wonder what's wrong with seb's waterfall then..
[19:49:47] <alex_joni> they are failing since 17 Nov
[19:50:36] <alex_joni> Runtest: 21 tests run, 5 successful, 15 failed + 1 expected
[19:51:03] <alex_joni> maybe some HAL stuff in use, and the HAL magic key changed..
[19:51:13] <alex_joni> I'll bug seb when he's around :)
[20:12:21] <cradek> jmkasunich: I'm getting the same joint-position-command jump for one servo cycle at index reset time that I got a while back when my addfs were in the wrong order.
[20:12:35] <cradek> however, I think they're right now, and it may be a problem in hm2. can you help me do a sanity check?
[20:16:34] <cradek> http://timeguy.com/cradek-files/emc/ferror-at-index.png
[20:18:16] <cradek> my threads are in this order: hm2-read, motion-command-handler, motion-controller, pid.0, pid.1, ..., hm2-write, hm2-pet
[20:33:08] <cradek> http://timeguy.com/cradek-files/emc/hm2-index-problem.png
[20:33:43] <cradek> this shows the hm2 pins. index-enable going false and the position reset do not seem to happen on the same cycle
[20:36:11] <cradek> in fact position jumps to an intermediate (?) value during that cycle when the index-enable goes low
[20:39:56] <cradek> yes I think the bug is in hm2_read
[20:40:24] <cradek> it calculates position (based on count) and then afterward checks the encoder and maybe resets index-enable and count
[20:41:16] <cradek> * cradek also waits for seb to show up
[21:03:46] <jmkasunich> cradek: I'm back now
[21:04:00] <cradek> hi
[21:04:13] <cradek> I'm > 50% convinced it's a hm2 bug
[21:04:18] <jmkasunich> I agree with your assesment - the bug is in the driver or firmware
[21:04:39] <cradek> ok, thanks
[21:04:48] <jmkasunich> when a canonical encoder clears index-enable, it MUST make the corresponding change in position on the same run
[21:05:24] <jmkasunich> that's the only way for EMC (or anything else) to deal with the jump
[21:05:26] <cradek> I think I showed in the second plot that it does not do that
[21:05:41] <jmkasunich> right - that pic is the key piece of data
[21:05:47] <cradek> but an ferror is happening too, which causes other kinds of resets
[21:06:16] <jmkasunich> can you turn the screws by hand?
[21:06:36] <cradek> with a bit of disassembly I could
[21:06:46] <jmkasunich> not that important
[21:06:50] <cradek> but that would prove it, wouldn't it
[21:06:56] <jmkasunich> a sure-fire test would be to leave emc out of the picture
[21:07:01] <jmkasunich> sets index-enable
[21:07:03] <jmkasunich> turn till index
[21:07:13] <cradek> ... look at plot
[21:07:16] <jmkasunich> with EMC and amps disabled
[21:07:37] <cradek> if seb doesn't see a problem right away, I will do that test
[21:08:03] <cradek> I think hm2_read() has the functions misordered
[21:08:18] <cradek> but he'll know for sure if that's the right fix
[21:08:35] <jmkasunich> I haven't looked at the code yet (just got back from a nice walk)
[21:08:50] <cradek> I shuld go outside before the sun sets - I think it's warm today...
[21:09:01] <cradek> in fact I'm going to do that. bbl!
[21:09:12] <jmkasunich> it was very pretty today, first blue sky in a week or so
[21:09:13] <jmkasunich> enjoy
[23:01:33] <alex_joni> jepler: not sure if you follow the rtai mailing list
[23:01:53] <alex_joni> a guy reports having problems with gcc 4.3.x and -O2