#emc-devel | Logs for 2008-10-02

[00:02:46] <CIA-40> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/drawing.c: change text for comparison/assignment block to be consistent with edit window
[01:15:24] <jmkasunich> regarding adaptive feed and the limitation of 0.0 to 1.0....
[01:16:05] <jmkasunich> if we allow >1.0, where do we draw the line? if someone asks for 1000, we're gonna be violating constraints all over the place
[01:16:36] <SWPadnos> or at least pegging to the vel/acc limits
[02:55:19] <cradek> like FO it wouldn't violate constraints
[03:08:33] <cradek> hm, got my nut back together, still .0012 backlash
[03:08:41] <cradek> much better than .0036 though
[03:08:59] <cradek> it might need more than reballing. but this makes it a lot more acceptable.
[03:09:25] <cradek> and, the broken balls are out, so maybe it won't deteriorate now
[03:11:10] <cradek> g0x0
[03:11:13] <cradek> oops
[03:18:14] <seb_kuzminsky> greetings
[03:20:13] <cradek> hi seb
[03:21:08] <seb_kuzminsky> cradek: have you tried hm2 pdm lately? ;-)
[03:21:58] <cradek> no, I'm happy with the old driver now, sorry to say
[03:22:18] <cradek> I'd like to switch sometime but I can't, without index mask
[03:22:23] <seb_kuzminsky> well it's good the old driver's working for you
[03:22:55] <cradek> yes it's working very well. jepler fixed the watchdog and added index mask for me.
[03:23:27] <cradek> well it was there, it was just hooked up to the wrong pins or something
[03:24:31] <seb_kuzminsky> for index mask you use an additional input pin to the quadrature counter? and sort of replace the old Index input with (Index && IndexMask)?
[03:25:21] <cradek> yes exactly
[03:25:57] <cradek> jepler brought that out to some inputs on the dio card
[03:26:25] <seb_kuzminsky> i think that will be easy to support in hm2, but first i'd really like to have some way to configure the driver that's better than the current modparam config string...
[03:26:41] <cradek> another approach would be to use an external AND gate... but ick.
[03:26:47] <seb_kuzminsky> iright
[03:26:48] <cradek> yeah that's sure complicated.
[03:27:08] <seb_kuzminsky> seems like it *should* be easy! just pass a structured object to the driver!
[03:27:14] <seb_kuzminsky> but nooooo
[03:27:36] <cradek> paul never came through with his suggestions did he. he seemed to think it was obvious but didn't care to share his thoughts.
[03:27:52] <seb_kuzminsky> sigh, paul...
[03:28:15] <seb_kuzminsky> one thing i was thinking about was to put all the config options in /proc or /sys or something, structured by dirs & files
[03:28:16] <cradek> "you're doing it stupidly" are a dime a dozen; "here's a great way" are priceless
[03:28:29] <seb_kuzminsky> let the users poke the right stuff into the config files there
[03:28:41] <cradek> interesting idea
[03:28:51] <seb_kuzminsky> then poke a "1" into the "go" file, and have it register everything with hal then
[03:29:20] <cradek> something in halcmd could do the poking for you?
[03:29:57] <seb_kuzminsky> "halcmd loadusr echo 1 > /proc/hm2/5i20/0/num_encoders"
[03:31:29] <cradek> halcmd sysctl -w ...
[03:31:36] <seb_kuzminsky> i think this sacrifices the nice feature that's there now, which is that if the config doesnt match the firmware, the insmod/loadrt fails
[03:32:37] <cradek> halcmd sysctl -p /etc/emc2/configs/.../my-firmware-config
[03:33:05] <seb_kuzminsky> nice
[03:33:31] <cradek> seems like sysctl gives you all the capabilities that you would probably otherwise go write
[03:33:46] <cradek> oh, it only works on /proc/sys/...
[03:33:49] <seb_kuzminsky> though there'd be some cryptic error message if the user-specified firmware doesnt match the user-specified sysctl config
[03:33:53] <cradek> that seems kind of stupid
[03:34:19] <cradek> seb_kuzminsky: yeah some later thing would fail I guess. those kinds of failures are always a pain.
[03:42:42] <cradek> I'm off to bed - goodnight
[03:42:50] <seb_kuzminsky> ok goodnight
[08:43:11] <micges> how can I measure time spent in HAL module main function ?
[08:43:30] <micges> rtapi_get_time or simmilar ?
[10:40:06] <CIA-40> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ladder/classic_ladder.lyx: update description of editor window to match and other edits
[10:42:26] <SWPadnos> micges, I think rtapi_get_time should work. if you're just monitoring, then there's already a parameter in HAL (called <function-name>.time)
[10:42:50] <SWPadnos> and <function-name>.tmax has the max time
[10:56:00] <micges> I want to add tmax to my function
[10:56:38] <SWPadnos> is this something that runs in a servo thread, while there's a base thread active?
[10:58:34] <micges> yes
[10:58:56] <micges> oh I see that there are time and tmax
[10:59:15] <micges> they are added automagically ?
[10:59:15] <alex_joni> remember they are in cpu cycles though
[10:59:19] <alex_joni> micges: yup
[11:00:38] <SWPadnos> they are also just (end_time-start_time), so if a base thread executes during execution, you'll get the whole base thread time (plus context switches)included
[11:02:32] <micges> I see why it oscillate so much
[11:37:29] <micges> there is no thread sync in hal ?
[11:38:00] <micges> base thread can be executed while servo-thread func is running ?
[11:38:21] <cradek> yes of course, maybe even several times
[11:38:41] <micges> that behavior isn't create bad behavior ?
[11:39:03] <cradek> on the contrary, I think the behavior is required
[11:39:15] <cradek> how else could you meet the timing requirement of the base thread?
[11:40:14] <alex_joni> if the base thread needs to be executed each 200 nsec, and the servo thread takes 1 msec to execute..
[11:40:14] <micges> right
[11:40:23] <alex_joni> you can't do it otherwise
[11:48:18] <SWPadnos> it does have the potential to cause problems, but it's unavoidable
[13:55:41] <SWPadnos> hey cradek, did you ever get one of those cheap motherboard/memory/CPU combos from NewEgg?
[13:55:50] <SWPadnos> (the Goal3+ board)
[14:03:58] <cradek> SWPadnos: no, I never really intended to get one, I just thought it was interesting.
[14:04:20] <SWPadnos> ok. I may get one soon - wasn't sure if anyone else had jumped on the bandwagon first
[14:04:31] <cradek> I have so much old hardware that I really never buy anything.
[14:04:34] <SWPadnos> heh
[14:04:40] <SWPadnos> but this would be new hardware! :)
[14:05:05] <cradek> often I find that old hardware is better...
[14:05:18] <SWPadnos> the nice thing about that board is that you can easily run it from a picoPSU, so you can have a system that runs off 12V or 24V DC
[14:05:25] <SWPadnos> yeah, especially if it's still running
[14:05:41] <cradek> that sounds nice
[14:05:50] <cradek> I had to put a whole PC supply in the lathe - it took up a lot of room.
[14:06:08] <SWPadnos> yeah. they're noisy and relatively large
[14:06:20] <SWPadnos> (electrical noisy at least)
[14:12:01] <skunkworks> I did buy a few of these for work.. http://www.newegg.com/Product/Product.aspx?Item=N82E16813135091
[14:12:10] <skunkworks> I have not tested realtime though
[14:12:35] <skunkworks> at the time they where 79.99
[14:12:45] <SWPadnos> yep, that's the one
[14:12:57] <SWPadnos> too bad you only get 1 rebate per address
[14:13:50] <skunkworks> when I bought them - (I think you pointed them out) they where 79.99 - no rebate.
[14:15:41] <skunkworks> I should power one up.
[14:16:00] <SWPadnos> yeah, there was a rebate when I first saw them, but I guess that was gone for a couple of weeks
[14:16:08] <SWPadnos> you must have gotten lucky ;)
[14:17:52] <skunkworks> aww shucks
[14:18:33] <skunkworks> yah - I paid $319.96 for 4
[14:18:46] <SWPadnos> how terrible! ;)
[14:55:16] <skunkworks> http://imagebin.org/27829
[14:56:18] <SWPadnos> cool. how's the latency? :)
[14:56:41] <skunkworks> not that far yet.. ;)_
[14:56:49] <SWPadnos> almost, it seems
[14:57:12] <skunkworks> I just went right to install...
[14:57:22] <skunkworks> rebooting right now.
[14:57:25] <cradek> what graphics chipset? hope it's the intel
[14:57:48] <SWPadnos> sis
[14:57:52] <SWPadnos> unichrome I think
[14:59:16] <skunkworks> heh - the manual says 'high performance and hight quality 3d graphics accellerator'
[14:59:34] <SWPadnos> I guess the authors were high
[15:01:11] <skunkworks> well - they did spell correctly though..
[15:02:46] <skunkworks> video is very noisy.. I don't know if it is the monitor or the fact I don't have it in a case.
[15:14:38] <skunkworks> http://imagebin.org/27832
[15:15:26] <skunkworks> you don't see the noise in the screenshot.. so maybe it is radiation.. gets worse when you run glxgears.
[15:16:35] <SWPadnos> hmmm. sounds like you have bad caps or something
[15:16:44] <SWPadnos> is it still there if you unplug the LCD?
[15:17:34] <skunkworks> no - visual noise
[15:17:52] <SWPadnos> ah, ok
[15:19:01] <skunkworks> up to 6456
[15:19:16] <SWPadnos> still excellent though
[15:19:44] <skunkworks> very
[15:20:04] <skunkworks> it locked up on me when I plugged in my usb drive.. but I might have zapped it. ;)
[15:20:56] <skunkworks> it has been running for 15min or more now.
[15:21:16] <skunkworks> next is to get the network setup and all updates and see what happens.
[15:22:16] <SWPadnos> well, I guess it's time for me to get a few then :)
[15:22:29] <SWPadnos> 8 USB ports too - could be useful for other stuff
[15:23:18] <skunkworks> yah - I just took it out again.. I have it setting I a large ungrounded metal case - I grounded myself on the case and the thing went dead.. ;)
[15:23:55] <SWPadnos> it's also on conductive anti-static packaging, which isn't ideal (especially if some pins punch through)
[15:24:38] <skunkworks> yes.
[15:43:04] <skunkworks> 262 updates
[18:03:50] <skunkworks> I think we have another winner http://www.electronicsam.com/images/KandT/testing/athlon64-3200ECSGoal3.png
[18:04:09] <alex_joni> wooot
[18:04:14] <alex_joni> is that using i386?
[18:04:36] <skunkworks> yes
[18:04:45] <alex_joni> maybe you can try amd64 if you feel like testing :P
[18:06:15] <alex_joni> and compare
[18:06:47] <skunkworks> is there a livecd setup for that? or can I just download the 64bit version of ubuntu and run the script?
[18:07:03] <alex_joni> the later
[18:07:15] <alex_joni> +t
[18:07:23] <alex_joni> brb
[18:07:39] <skunkworks> ok - hmm - I wonder if I have it downloaded already...
[18:08:16] <alex_joni> I bet you'll find it _after_ you finished the new download
[18:08:51] <skunkworks> heh
[18:09:06] <skunkworks> hmm - I have the beta
[18:09:12] <alex_joni> good enough
[18:09:17] <skunkworks> ok
[18:09:22] <alex_joni> lots of updates though
[18:09:31] <skunkworks> * skunkworks gets a blank cd
[18:53:13] <skunkworks> huh - my beta version has to do a 'package upgrade'
[18:54:14] <skunkworks> distribution package upgrade
[20:06:57] <alex_joni> hi seb_kuzminsky
[20:08:47] <seb_kuzminsky> hi alex_joni
[20:21:37] <micges> is there nml message to change vel/acc of axis ?
[20:22:51] <micges> or this is setable directly by ini ?
[20:23:11] <SWPadnos> I think there's a message to init an axis, which means to read the config from the ini file
[20:56:23] <skunkworks> how do I know if I have the 64bit kernel
[20:57:30] <jepler> skunkworks: 'uname -m' will print x86_64 instead of i686
[20:59:36] <skunkworks> is there a separate install script for 64?
[21:00:45] <jepler> no, I think it's the same install script
[21:02:28] <skunkworks> hmm - then I think I may need to just install from the livecd. I installed the only copy I had which was hardy64 beta. This then updated but said something about not being able to udate all 'partial update'. after that it said it was up-to-date. I ran the install script and rebooted again - latency-test doesn't run.
[21:04:19] <skunkworks> heh - wrong kernel
[21:04:23] <skunkworks> duh
[21:05:49] <skunkworks> nope - still doesn't work.
[21:05:51] <micges> what does EMC_TRAJ_SET_ACCELERATION message do ?
[21:06:06] <skunkworks> I think this is borked. I have the latest 64bit version..
[21:07:36] <jepler> skunkworks: hm, uh oh
[21:07:38] <jepler> what's in dmesg?
[21:08:02] <micges> there is EMC_AXIS_SET_MAX_VELOCITY message but there is no acc related msg
[21:09:34] <skunkworks> jepler: thanks - but I am installing the latest 64 and go from there. the beta version I had was from quite a while ago.
[21:16:40] <SWPadnos> eh, we're only ~3 weeks from 8.10, so why bother? :)
[21:17:39] <alex_joni> SWPadnos: planning on supporting 8.10 ?
[21:17:49] <SWPadnos> err, well no, not me personally
[21:17:59] <SWPadnos> * SWPadnos whistles away quietly
[21:18:01] <alex_joni> hmm.. maybe PC will
[21:18:17] <SWPadnos> ok, I've just started holding my breath
[21:18:22] <SWPadnos> let me know when to stop
[21:18:30] <alex_joni> you can stop.. it's pointless
[21:18:36] <SWPadnos> phew
[21:18:49] <alex_joni> meh.. kitchens are really expensive :/
[21:19:00] <jepler> kittens? I love kittens!
[21:19:34] <cradek> http://www.dailykitten.com/archives/2181-minion.html
[21:19:39] <cradek> Minion! what a great name!
[21:19:54] <jepler> indeed it is
[21:20:24] <cradek> uh, it's 4:20, and I'm still at work
[21:20:29] <cradek> what am I thinking!?
[21:20:38] <alex_joni> "I should go home.."
[21:20:48] <alex_joni> * alex_joni is a world famous mind reader
[21:36:21] <skunkworks> * skunkworks reads the latest email on the list..
[21:36:23] <skunkworks> ;)
[21:40:07] <alex_joni> I'd suspect a memlock issue (out of the blue..)
[21:41:21] <skunkworks> heh - mine only has 512mb. But I am installing and updating the latest ubuntu64
[21:41:34] <skunkworks> I will let you guys know.
[21:42:31] <skunkworks> probably tomorrow
[21:43:39] <alex_joni> I meant something else with memlock :P
[23:57:46] <CIA-40> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/edit_gtk.c: change editor window tooltips text for object selector