#emc | Logs for 2006-03-02

[00:01:08] <skunkworks> now that you have your head around the tp - would it be possible with the current setup to do "tollernce blending" don't really know what it would be called. where you tell it you want the path to be within a certain tollerance - lets say .010". It would adjust the feed accordingly to make the correct path.
[00:01:30] <cradek> yes I think so.
[00:02:05] <skunkworks> nice.
[00:02:10] <cradek> my current algorithm is "blend as much as possible while still touching at least part of every segment" which as you saw in my photo the old TP doesn't do
[00:02:31] <cradek> photo = screen grab
[00:02:37] <skunkworks> right
[00:03:18] <cradek> in that program, it will slow down markedly for those short segments at the ends, but that's necessary because it's reversing anyway. might as well stay on the path.
[00:03:27] <skunkworks> that will probably work for 99 percent of the people out there.
[00:04:59] <skunkworks> very good work. when I had made a simple motion controller in basic - I had exact stop working and started to think about read ahead and blending. that is were I quit ;). I could not get my head around it.
[00:05:55] <skunkworks> then I found turbocnc which did everything my program did plus a billion times better.
[00:06:14] <cradek> well don't say good work until it works... I'm getting a blend in the right direction now, but it's bogus...
[00:06:46] <skunkworks> you will get it.
[00:07:01] <cradek> probably true. I've made a lot of progress today.
[00:07:48] <skunkworks> wish I could help but about all I can do is test it when it is to that point.
[00:08:13] <cradek> that's fine, there are enough cooks on this soup already (1)
[00:08:20] <cradek> hmm, I'm hungry
[00:08:25] <cradek> haha
[00:08:27] <skunkworks> :)
[00:09:06] <skunkworks> I am glad you have help.
[00:11:30] <cradek> wooo
[00:11:43] <cradek> haha you have to see this one
[00:11:56] <skunkworks> not quite what you had in mind?
[00:12:06] <cradek> http://timeguy.com/cradek-files/emc/not-quite.png
[00:12:17] <cradek> but look at that shape
[00:12:53] <jepler> if only it was in the right place
[00:12:58] <jepler> this is simple_tp ?
[00:13:01] <skunkworks> a little off but the shape looks good
[00:13:01] <cradek> yes
[00:13:05] <jepler> neat
[00:13:25] <cradek> jepler: http://timeguy.com/cradek-files/emc/nasty-blend.png
[00:13:31] <cradek> this is the current behavior
[00:13:42] <cradek> huuungry
[00:13:43] <cradek> bbl
[00:21:01] <jepler> current meaning HEAD tp?
[00:22:33] <skunkworks> pretty sure that is what he means.
[00:28:39] <cradek> yes HEAD
[00:29:12] <cradek> (running with very low accel makes it worse)
[00:35:07] <cradek> http://timeguy.com/cradek-files/emc/whee.png
[00:42:33] <skunkworks> very nice.
[00:42:36] <skunkworks> bbl
[01:13:07] <SWPadnos> nice whee there, cradek
[01:13:21] <cradek> thanks
[01:13:24] <cradek> this is extremely promising
[01:13:32] <SWPadnos> it certainly is
[01:13:51] <SWPadnos> the yellow is because the TP put in an arc?
[01:14:14] <SWPadnos> hmmm - no, that makes no sense, it's actually an elliptical section
[01:14:22] <cradek> not really - the yellow is a hack that allows me to see the blends
[01:14:29] <SWPadnos> heh - OK. :)
[01:15:15] <cradek> http://timeguy.com/cradek-files/emc/whee2.png
[01:15:27] <cradek> this shows line-line line-arc and arc-line
[01:15:48] <cradek> it's hard to see but the magenta means an arc is active
[01:15:53] <SWPadnos> right - the fuschia is a programmed arc?
[01:15:58] <cradek> yes
[01:15:58] <SWPadnos> oops - magenta
[01:16:07] <cradek> we can blame jepler for the colors :-)
[01:16:08] <SWPadnos> I need to calibrate this montior
[01:16:10] <SWPadnos> ok
[01:16:20] <SWPadnos> can this mode be left in as an option in axis?
[01:16:34] <SWPadnos> "debug plot" mode
[01:16:50] <cradek> well axis doesn't know a blend from its rear end - it's in the tp output
[01:17:14] <cradek> I have the tp tell axis it's doing a jog when in fact it's blending
[01:17:26] <SWPadnos> ah, so yellow is the jog color
[01:17:29] <cradek> yes
[01:17:51] <SWPadnos> ok. it should be possible to add "blend" and any other segment types we can thinkn of though
[01:17:55] <cradek> but we could always add another color: jog, feed, arc, blend
[01:18:08] <cradek> yes
[01:18:14] <SWPadnos> you have marks for dwell, right?
[01:18:41] <SWPadnos> or were those 'x'es in the sample screenshots for peck cycles?
[01:18:44] <cradek> yes little Xs
[01:19:07] <cradek> http://timeguy.com/cradek-files/emc/spot-the-bug.png
[01:19:10] <SWPadnos> and a separate segment type for g0, right?
[01:19:26] <cradek> yes g0 is grayish
[01:19:55] <SWPadnos> no g0 vertical on the right side plunge
[01:19:58] <cradek> feed->jog blend is not quite right (pretty funny)
[01:20:08] <cradek> yeah see all the yellow? shouldn't be there
[01:20:20] <cradek> skunkworks: http://timeguy.com/cradek-files/emc/whee2.png
[01:20:22] <SWPadnos> the long horizontal at the top
[01:20:28] <cradek> SWPadnos: yep
[01:21:05] <SWPadnos> blending of g0+G[12] is a question
[01:21:10] <skunkworks> wow - nice job. any issues?
[01:21:22] <cradek> skunkworks: one obvious bug left: http://timeguy.com/cradek-files/emc/spot-the-bug.png
[01:21:55] <cradek> but I think to my amazement I'm nearing a checking for people to play with
[01:21:59] <SWPadnos> all we have to add is an image of the block of material being milled away, and axis will be *really* cool! :)
[01:22:07] <cradek> "we"??
[01:22:11] <SWPadnos> err - jepler
[01:22:15] <cradek> ok
[01:22:17] <SWPadnos> heh
[01:22:36] <skunkworks> and a tool shape file ;)\
[01:22:52] <SWPadnos> of course, that's needed for proper material removal renderings
[01:23:05] <cradek> and the material should glow red if you cut too fast
[01:23:13] <SWPadnos> and smoke if you linger too long
[01:23:23] <skunkworks> we should have chip rendering also
[01:23:42] <cradek> and the tool randomly breaking off if it's 1/16th or smaller
[01:23:43] <SWPadnos> and the tool should break in half and fly away if the spindle load suddenly decreases (unless you're at the end of the material)
[01:24:00] <cradek> ok, that's better than my idea
[01:24:04] <SWPadnos> then it would be *really really* cool!
[01:24:35] <cradek> ooh, found arc-arc blend on cds, it's also right
[01:25:01] <skunkworks> I think it is already *really really* cool! we need to add some reallys
[01:25:34] <SWPadnos> ok, all those eye-candy features would be *really really really really really* cool!!!
[01:25:46] <SWPadnos> (extra exclamation points too)
[01:26:01] <skunkworks> really^10 cool!^40
[01:26:31] <SWPadnos> ^2
[01:26:32] <skunkworks> sorry really^10 cool(!^40)
[01:26:57] <SWPadnos> (*really*)^10 cool(!^40)
[01:27:01] <SWPadnos> ( *really* )^10 cool(!^40)
[01:27:03] <SWPadnos> there
[01:27:33] <skunkworks> ok - now that is just goofy
[01:27:50] <SWPadnos> heh
[01:36:15] <cradek> there, fixed the g1->g0 bug
[01:36:57] <jepler> cradek: I'm glad somebody's finally taken simple_tp and "run with it". If not for your groundwork, it never would have happened.
[01:37:22] <skunkworks> here here
[01:37:27] <cradek> I appreciate that - I didn't think I'd be the one to continue it...
[01:37:28] <SWPadnos> and here
[01:37:39] <SWPadnos> heh
[01:39:59] <cradek> now I'll feed the output to stepgen to see if I'm really obeying constraints...
[01:41:58] <cradek> http://timeguy.com/cradek-files/emc/garish-whee3.png
[01:42:11] <cradek> the colors are evil but it sure looks right
[01:43:17] <SWPadnos> very pretty
[01:43:42] <skunkworks> nice
[01:44:52] <cradek> hmm
[01:44:53] <cradek> emc/task/taskintf.cc 787: Error on axis 2, command number 107
[01:45:01] <cradek> anyone know what this means?
[01:45:18] <SWPadnos> cradek, how hard do you suppose it would be to add some parameters to the TP that control the blend mode, whether or not to blend g0 with g1 or g2, etc?
[01:45:28] <SWPadnos> not off the top of my head
[01:45:36] <cradek> SWPadnos: SMOP
[01:45:46] <SWPadnos> ?
[01:45:49] <cradek> sorry
[01:45:53] <cradek> "simple matter of programming"
[01:45:57] <SWPadnos> ok :)
[01:46:07] <cradek> a friend of mine used to say that when something was techinically possible, no matter how hard
[01:46:13] <SWPadnos> heh
[01:46:22] <SWPadnos> it's pretty close to SEP, so I was wondering ;)
[01:46:30] <cradek> what's that?
[01:46:38] <SWPadnos> "Somebody Else's Problem"
[01:46:43] <cradek> ah
[01:47:20] <cradek> once this is polished, I can probably add pretty much to the TP as long as it doesn't take any more complex math than what's here
[01:47:28] <cradek> ... pretty much anything
[01:48:03] <SWPadnos> yep - or even if it does
[01:48:22] <SWPadnos> I was thinking that some of these options shouldn't be decided at programming or compile time, they should be settable modes
[01:48:49] <SWPadnos> options like how to deal with an empty queue, how to blend certain types of move combinations, etc
[01:48:51] <cradek> yeah, getting settings from the ini file through all the layers of crap is probably the only hard part.
[01:49:04] <SWPadnos> HAL pin (if there's anywhere it could go)
[01:49:29] <cradek> I like fenn's suggestion about a g code that specifies "don't blend turns greater than n degrees"
[01:49:42] <SWPadnos> weird - simple_tp isn't returned in the list of branch tags
[01:49:45] <cradek> I think it was fenn
[01:49:47] <SWPadnos> yep - that's a good one too
[01:49:51] <cradek> it's only a branch in the kinematics directory
[01:50:02] <SWPadnos> maybe jmk - he was against an arbitrary choice of 90 deg
[01:50:05] <SWPadnos> ah, right
[01:53:11] <cradek> whee, constraints look right running cds
[01:53:42] <SWPadnos> cool
[01:53:52] <SWPadnos> that error looks like a catch-all
[01:54:05] <SWPadnos> if there's an error, and it's not an RCS error, print that message
[01:54:40] <cradek> emc/task/taskintf.cc 787: Error on axis 2, command number 172
[01:54:50] <cradek> I think this should have been a following error - not sure why it wasn't
[01:56:17] <SWPadnos> that particular routine doesn't check for following errors
[01:59:23] <jepler> INTERP_MIN_ERROR
[01:59:24] <jepler> is therefore the index of the last non-error return value.
[01:59:32] <jepler> </a header file>
[01:59:38] <jepler> so INTERP_MIN_ERROR is not, in fact, an error?
[02:01:21] <CIA-8> 03cradek 07stp_trap_blend * 10emc2/src/emc/kinematics/ (tc.h tp.c):
[02:01:21] <CIA-8> Spectacular initial results for a smarter trapezoidal blending based
[02:01:21] <CIA-8> on simple_tp. A relatively up-to-date AXIS will display blends in
[02:01:21] <CIA-8> yellow so you can more easily see what's going on.
[02:02:30] <SWPadnos> oooohh - gotta go, there's a Monty Python special on PBS :)
[02:10:56] <jepler> cradek: (neat^3)^17
[02:15:05] <cradek> jepler: the code's not pretty yet, but it works and is understandable
[02:40:31] <jepler> if self.state.feed_mode <= 30:
[02:40:59] <jepler> any idea what this is testing for? What is feed_mode? apparently it's called gcodes[7] by the rs274ngc interp
[02:41:33] <cradek> sorry, no clue
[02:42:10] <jepler> group 5 - gez[7] g93, g94 - feed rate mode
[02:42:23] <jepler> What in the world would a dwell mean in inverse time feed mode?
[02:42:52] <jepler> (and if that's what I meant to test for, I wonder if it's the right test)
[04:37:23] <CIA-8> 03cradek 07stp_trap_blend * 10emc2/src/emc/kinematics/tp.c: (log message trimmed)
[04:37:23] <CIA-8> Fix one last accel constraint violation that showed up on reversals
[04:37:23] <CIA-8> (acute angles) Also add 95% hack to be nice to stepgen and no longer
[04:37:23] <CIA-8> require the user set up stepgen headroom (it's still needed for
[04:37:23] <CIA-8> jogs unless I can talk jmk into doing the same thing in motion).
[04:37:23] <CIA-8> With this version I have really excellent blending results running
[04:37:27] <CIA-8> CDS on Ray's 3ips stepper ini with feed override 900%. I'm pretty
[05:11:05] <dave-e> SWP??
[05:11:26] <SWPadnos> err - I'm asleep ;)
[05:11:36] <dave-e> or almost?
[05:11:43] <SWPadnos> heh - what's up?
[05:12:00] <dave-e> had a bad day...Y encoder input quit
[05:12:11] <SWPadnos> bummer - that'll cause some trouble
[05:12:28] <dave-e> but have been working on X...
[05:12:46] <dave-e> OUTPUT_SCALE seems backwards
[05:13:04] <SWPadnos> with the old driver, or with jmk's change?
[05:13:04] <dave-e> lower values give higher output
[05:13:12] <dave-e> old driver...
[05:13:51] <dave-e> I think chris was going to stick something together for me but haven't heard a peep
[05:14:19] <SWPadnos> odd, the driver does multiply, so larger numbers should give higher output
[05:14:31] <dave-e> but somehow Xoutput seems clamped at 3.25
[05:15:00] <SWPadnos> IPS or volts?
[05:15:03] <dave-e> so if I demand much the system fails on following error
[05:15:16] <dave-e> volts
[05:15:53] <SWPadnos> pid.n.maxoutput is set to MAX_VELOCITY from the ini file
[05:15:58] <dave-e> maybe I should wait for updates
[05:16:00] <SWPadnos> is that 3.25?
[05:16:24] <dave-e> it is ... am I looking at the wrong thing
[05:16:33] <SWPadnos> could be ;)
[05:17:02] <SWPadnos> the planner should never ask for more than MAX_VEL, but the PID is guaranteed to never exceed it
[05:17:39] <SWPadnos> so if the TP does ask for more, or there's roundoff or the like, you can get an ferror
[05:19:57] <dave-e> no I'm crashing early....depending on OUTPUT_SCALE values near on work pretty welll so far... .3 gets pretty hot .. and 3.25 causes doesnt/ work either
[05:20:06] <dave-e> near one
[05:20:25] <SWPadnos> that's weird
[05:20:47] <SWPadnos> have you tried jmk's suggestion of running just HAL, no EMC?
[05:20:50] <dave-e> this is backwards of what jmk said last nite.....and may be related to the units thing
[05:21:31] <dave-e> not really.... I don't navigate the system very well.
[05:21:45] <SWPadnos> I'm looking at code from before his changes
[05:22:08] <dave-e> tried to bring up halscope tonight and it kept telling me the scope_rt.o didn't exist
[05:22:14] <dave-e> that
[05:22:43] <SWPadnos> how were you trying to load it?
[05:22:58] <dave-e> insmod scope_rt.o
[05:23:11] <dave-e> what should I be doing
[05:23:15] <SWPadnos> try using halcmd instead, it knows where to look for it, and it knows what extension to use
[05:23:23] <SWPadnos> halcmd loadrt scope_rt
[05:23:45] <dave-e> certainly worth a try
[05:24:05] <SWPadnos> I'd definitely try a HAL-only test
[05:24:08] <dave-e> I'd like to get one axis happy before I send the board off for repair
[05:24:48] <SWPadnos> I see that the driver does do some strange math (((-10 - volts)/20) * 0x1FFF)
[05:25:18] <dave-e> offset-volts/span
[05:25:24] <SWPadnos> yep, but negative
[05:25:30] <SWPadnos> -10 - volts
[05:25:44] <dave-e> hmmm
[05:25:53] <SWPadnos> volts is calculated as (value - offset) * scale
[05:25:59] <SWPadnos> which makes sense
[05:26:24] <dave-e> then time the hex?
[05:26:30] <dave-e> times
[05:26:38] <SWPadnos> right - to convert to counts
[05:27:17] <dave-e> zero should be FFF?
[05:27:43] <SWPadnos> I guess so, and -10 = 0x1FFF
[05:27:53] <SWPadnos> +10 = 0
[05:28:02] <SWPadnos> oops, -10 = 0 counts
[05:28:30] <SWPadnos> I'm not sure how the hardware actually works though, so that could be 2's complement, or sign + value
[05:28:44] <dave-e> I'd have to look at the manual...some dacs are different than others
[05:28:52] <SWPadnos> yep
[05:30:00] <Jymmm> Hey, who had the cnc etchasketch?
[05:30:05] <SWPadnos> jepler
[05:30:12] <Jymmm> thanks
[05:30:19] <SWPadnos> sure
[05:31:23] <dave-e> If I download the emc cvs how much trouble am I going to have compiling?
[05:31:57] <SWPadnos> that depends on whether or not you've ever compiled emc on that machine
[05:32:11] <dave-e> not on that machine
[05:32:13] <SWPadnos> what distro do you have?
[05:32:23] <dave-e> ubuntu
[05:32:27] <dave-e> 5.10
[05:32:33] <SWPadnos> on the emc machine?
[05:32:37] <dave-e> yes
[05:33:03] <SWPadnos> ok. bring it into the house (or get a very long network cable), and do "apt-get build-dep emc2"
[05:33:42] <dave-e> I should be myself a 100' patch cable
[05:34:02] <dave-e> or better yet run about 200' of cable in conduit
[05:34:03] <SWPadnos> you definitely should be
[05:34:27] <SWPadnos> that would also work, but you'd want to be sure that you only use a switch on that segment, not a hub
[05:34:34] <dave-e> it is a pain to dismount the computer from the cabinet
[05:35:08] <SWPadnos> ok. long cord it is. Home Depot has 1000 foot spools for ~$55
[05:35:19] <dave-e> 200' should not be a problem the spec is 105 m or so
[05:35:56] <SWPadnos> 100m hub to node, 5m between hubs, max 2 hubs on the same collision domain, IIRC
[05:36:23] <SWPadnos> <512 bit times total end-to-end "electrical distance
[05:36:26] <SWPadnos> "
[05:37:24] <dave-e> you mean 2 deep?
[05:37:49] <SWPadnos> node -- 100m cable -- hub -- 5m cable -- hub -- 100m cable -- node
[05:38:04] <dave-e> OK
[05:38:34] <SWPadnos> add a switch, and you can repeat the whole thing, putting the switch at one of the "node" points
[05:39:08] <dave-e> I keep thinking I'll get the fiber installed...I have a few Km of plenum fiber 62.5/125
[05:39:31] <dave-e> but my drivers are AUI not 10baseT
[05:39:35] <SWPadnos> hmmm - that can work for 10gigE, I think (and possibly infiniband and/or SONET)
[05:40:14] <dave-e> 10 meg will do just fine for the traffic I have
[05:40:41] <SWPadnos> I'm probably thinking of other uses
[05:40:45] <SWPadnos> (not pr0n)
[05:41:03] <dave-e> eh?
[05:41:16] <SWPadnos> high speed camera array communications
[05:41:38] <dave-e> there bandwidth is just fine
[05:42:17] <dave-e> I think the fiber is 160 MHz/Km so for short distances it should really scream
[05:42:30] <SWPadnos> I was looking into InfiniBand recently, and I thought the specs were for multimode fiber, similar to the wavelengths you mentioned
[05:42:51] <dave-e> hmmm
[05:43:00] <SWPadnos> there may be more stringent specs for insertion loss and the like though
[05:44:14] <dave-e> but you generally have an energy budget of 8 db or so
[05:45:00] <SWPadnos> good to know. the distances I'll need are relatively short for fiber as well - maybe 500m
[05:45:29] <dave-e> the connectors should not lose much...and if I got desperate I'd try immersion oil to mask the interface
[05:45:58] <SWPadnos> does that stick when you move stuff around?
[05:46:12] <SWPadnos> (ie, take the whole system, move it to another town, and set it all up again)
[05:46:14] <dave-e> it is pretty viscous
[05:46:50] <dave-e> as long as the connectors don't flex within themselves it shoud be ok
[05:47:00] <SWPadnos> cool - I'll have to write that down
[05:47:59] <dave-e> it is used on microscopes to give better resolution at high power....940X or twice the hi dri objective
[05:48:51] <SWPadnos> ok. I know there are lenses and the like that use an oil layer to improve various optical properties - hadn't heard of cipping fiber in oil though :)
[05:48:52] <dave-e> goes between the objective and the microscope slide
[05:48:59] <SWPadnos> dipping
[05:49:15] <dave-e> don't see why it won't work
[05:49:24] <dave-e> maybe the gain isn't worth it.
[05:49:27] <SWPadnos> me either
[05:49:53] <SWPadnos> I'd imagine that it's better for semi-permanent connections though, I'd bet that dust would stick to the oil quite nicely
[05:49:58] <dave-e> if you were really tight an gained .5 db on each end...
[05:50:12] <dave-e> dust will make a real mess.
[05:50:25] <dave-e> you want to stick the stuff together and leave it.
[05:50:50] <SWPadnos> no can do with this system - it has to go to rodeos and the like
[05:51:13] <SWPadnos> which can get dusty, I hear
[05:51:35] <dave-e> hmmm....maybe go with the mil-spec battlefield fiber you can drive over
[05:51:55] <SWPadnos> well, failure is an option, if cost is taken into consideration ;)
[05:52:25] <SWPadnos> but there's probably going to be a budget of a couple $k for cables, so it should be covered
[05:53:31] <dave-e> can't remember the name of the outfit...IIRC south of Washington DC...on the coast
[05:54:24] <SWPadnos> hmmm. I should look into some of the outfits that make kevlar cable with other stuff in it (like fiber / copper, etc)
[05:54:51] <dave-e> If I stumble across a name I'll relay to you
[05:54:57] <dave-e> ....brb
[05:55:00] <SWPadnos> cool. thanks
[05:55:12] <SWPadnos> ok. I may be gone, it's past my bedtime
[05:57:51] <dave-e> optical cable corp 1-800-622-711... may be an old number Roanoke, VA
[05:58:12] <dave-e> 622-7111
[05:58:30] <SWPadnos> thanks - I should be able to google for them
[05:58:35] <dave-e> third times a charm....1-800-622-7711
[05:58:42] <SWPadnos> heh
[05:58:54] <dave-e> good luck...see ya later.
[05:59:00] <SWPadnos> thanks - see you
[06:26:55] <SWPadnos> SWPadnos is now known as SWP_Away
[12:06:12] <cncuser> hello
[12:23:18] <alex_joni> hello
[12:37:50] <alex_joni> cncuser: http://video.google.com/videoplay?docid=-5284203655121476073
[12:37:57] <cncuser> hi alex
[12:45:02] <cncuser> my english is way better ;)
[12:47:39] <jepler> that's pretty funny
[12:48:04] <alex_joni> I am sinking...
[12:48:18] <alex_joni> hi jeff
[12:49:10] <cncuser> jo jepler :)
[13:03:45] <alex_joni> cncuser: http://video.google.com/videoplay?docid=-2512249746173560379
[13:10:39] <alex_joni> ouch.. don't drive around planes http://video.google.com/videoplay?docid=-1309610693318372088
[13:13:54] <cncuser> hmm, what should be /usr/local/bin/axis ?
[13:18:01] <cncuser> if i launch a make installed emc and change the DISPLAY to axis. emc searhces for that... what should this be ? a link to axis.py ?
[13:18:15] <cncuser> emc2 ;)
[13:21:27] <cncuser> ah, ok, got it
[13:21:34] <alex_joni> no, if you install axis it should get installed in /usr/bin/axis
[13:21:35] <alex_joni> afaik
[13:21:43] <cncuser> hmm, it doenst with me
[13:21:48] <cncuser> but no problem
[13:21:54] <cncuser> its a older version
[13:22:14] <alex_joni> you need to get a newer one
[13:22:24] <cncuser> hmm
[13:22:32] <cncuser> i have issues with compiling
[13:22:38] <cncuser> i dont want to break the thing again
[13:22:43] <cncuser> i need to build a iso
[13:23:08] <jepler> If you're using emc2 without --run-in-place, then axis is installed to the --prefix= used when you built Python
[13:23:13] <cncuser> for example latley, i just cant get emc to compile :) it allways loops on the depends...
[13:23:18] <jepler> so if python is in /usr/local/bin then axis should be too
[13:23:24] <cncuser> jepler: ill copy it over
[13:24:59] <jepler> "should be" means "that's what I think setup.py will do"
[13:25:39] <cncuser> yes. its a very unclean development environment here ;)
[13:26:10] <jepler> If the emc2 build "loops", check whether you have a file that is dated in the future.
[13:26:42] <cncuser> ok
[13:26:48] <cncuser> thats a possibility
[13:27:08] <jepler> though in that case I'd expect a warning similar to
[13:27:08] <jepler> make: Warning: File `/home/emc/emc2/src/Makefile.inc' has modification time 5.9e+04 s in the future
[13:27:28] <cncuser> hmm, no, just endless looping
[13:29:41] <cncuser> EMCROOT=/usr/src/emc2testing python setup.py install --force
[13:29:45] <cncuser> this should do it ?
[13:29:51] <cncuser> or is emc root /usr ?
[13:30:03] <cncuser> for it spreads a lot in /usr ?
[13:31:04] <jepler> export EMCROOT := /usr
[13:31:11] <jepler> the debian package uses EMCROOT=/usr
[13:31:17] <cncuser> ok
[13:31:27] <jepler> it is for emc2 configured with --prefix=/usr too
[13:31:44] <cncuser> i must say, i dont like the way emc gets hacked into pieces and spreaded all over /usr
[13:31:59] <jepler> run_installed = (os.environ.has_key("EMC_RUN_INSTALLED") \
[13:32:12] <jepler> I think you also have to set this environment variable to get the axis script in the right place
[13:32:21] <cncuser> but i go with it, as for i want as little "spezialthings" as possible
[13:32:25] <jepler> cncuser: Then use ./configure --enable-run-in-place && make && sudo make setuid
[13:32:32] <cncuser> jepler: i know, but emc2 got thatw ay, i follow
[13:32:35] <cncuser> i know ;)
[13:32:50] <cncuser> i could also install it to /moon
[13:33:27] <cncuser> .. /usr/local/etc is one of the wackiest things
[13:36:01] <cncuser> ihmmm
[13:36:53] <cncuser> hmm, no
[13:37:49] <cncuser> doesnt work
[13:38:21] <cncuser> # EMCROOT=/usr python setup.py install --force
[13:38:22] <cncuser> Traceback (most recent call last):
[13:38:22] <cncuser> File "setup.py", line 132, in ?
[13:38:22] <cncuser> emcplat = os.getenv("PLAT", find_emc_plat(emcroot))
[13:38:22] <cncuser> File "setup/emc_setup.py", line 50, in find_emc_plat
[13:38:22] <cncuser> for plat in ['nonrealtime'] + os.listdir(emcplatdir):
[13:38:23] <cncuser> OSError: [Errno 2] No such file or directory: '/usr/emc/plat
[13:38:30] <cncuser> and, yesits not here for its in the sourcedir
[13:38:36] <cncuser> so i dont know now ?
[13:38:45] <cncuser> should it point to the installdir or sourcedir ?
[13:38:48] <jepler> setup.py didn't detect the presence of emc2
[13:39:22] <jepler> it expects one of these files to be present:
[13:39:23] <jepler> emc2_marker = os.path.join(emcroot, "include", "config.h")
[13:39:23] <jepler> emc2_marker2 = os.path.join(emcroot, "include", "emc2", "config.h")
[13:39:23] <jepler> is_emc2 = os.path.exists(emc2_marker) or os.path.exists(emc2_marker2)
[13:39:31] <jepler> /usr/include/config.h or /usr/include/emc2/config.h
[13:39:37] <jepler> that's installed by 'make install'
[13:39:40] <jepler> of emc2
[13:40:00] <cncuser> hmmm
[13:40:02] <jepler> (the first case is for the --enable-run-in-place/classic setup, and the second is for the new default to run installed)
[13:40:39] <cncuser> hmm
[13:40:42] <cncuser> got nothing of that
[13:40:44] <cncuser> :(
[13:40:49] <cncuser> damn
[13:40:57] <cncuser> too many things changed in the last 2 months
[13:41:20] <cncuser> my whole howto has holes because the world keeps spinning around :(
[13:42:04] <cncuser> i dont know... should i whipe emc2 and axis off and install it again.. or should i just copy the things together i need
[13:42:31] <cncuser> if the firt would cost me more then a hour id throw my pc out the window... thats why i dont want to risk this
[13:44:20] <alex_joni> hmm.. I think you need thicker windows
[13:45:05] <cncuser> ;)
[13:45:12] <jepler> developing on Linux can never pi** me off so much as trying to develop on windows
[13:46:23] <jepler> in fact windows is a thorn in my side at this very moment
[13:46:35] <jepler> so I don't have a lot of sympathy for anyone else
[13:48:15] <alex_joni> no, I meant proper windows
[13:48:18] <jepler> none to spare, I'm too busy feeling sorry for myself
[13:48:28] <jepler> alex_joni: I understand that
[13:48:35] <alex_joni> cncuser said he wants to throw his pc out the window
[13:48:40] <jepler> alex_joni: I understand that
[13:48:44] <jepler> but I want to talk about my own problems!
[13:48:47] <alex_joni> jepler: then I'm lost..
[13:49:16] <jepler> alex_joni: I'm talking about something that is going on at my job right now.
[13:49:26] <alex_joni> oh.. ok ;)
[13:50:33] <ValarQ> mr Kasunich sure loves comments
[13:51:39] <alex_joni> ValarQ: talking about what?
[13:51:40] <cncuser> well, drop the job, burn the windowsbox :) party :)
[13:51:56] <cncuser> easy as cake
[13:51:57] <alex_joni> jepler: what problems?
[13:51:57] <ValarQ> alex_joni: i'm reading stepgen.c
[13:52:02] <alex_joni> ValarQ: heh ;)
[13:52:07] <jepler> alex_joni: Oh, I don't want to talk about the details.
[13:52:20] <cncuser> jepler: dont like my solution ?
[13:52:30] <jepler> cncuser: it would kinda make my employer unhappy
[13:53:24] <cncuser> jepler: well, its either your employer or you that is happy
[13:53:27] <alex_joni> jepler: I understand..
[13:54:58] <cncuser> back in the days we settled this by creating mana 50 firethundermagic and afterwards got drunk at the 'loose cows In'
[13:57:24] <cncuser> ok, no more spellcasting
[13:57:27] <cncuser> ;)
[13:59:08] <cncuser> jepler: maybe it would cheer you up if you could choose the name for the new puppysit rellease, first time in history featuring axis on a livecd .. afaik ?
[14:00:08] <cncuser> jepler: for you did most of the job, fighting the python monster
[14:01:01] <cncuser> its a little big. 49mb. but still runs with 128mb ram
[14:07:40] <skunkworks> alex - did you see what cradek did with the simple tp last night
[14:08:31] <alex_joni> slightly
[14:10:04] <skunkworks> this http://timeguy.com/cradek-files/emc/nasty-blend.png
[14:10:29] <skunkworks> to this http://timeguy.com/cradek-files/emc/whee2.png
[14:10:42] <jepler> nasty-blend is with the HEAD tp
[14:11:55] <jepler> "His project is lighting up 18 LEDs using one resistor."
[14:12:10] <jepler> wow, I wonder how he does it without a power source too
[14:12:42] <skunkworks> was that ment for here? ;)
[14:13:26] <cncuser> driving leds on AC ?
[14:13:29] <cncuser> cool
[14:13:48] <cncuser> would be a nize 50Hz pulse :)
[14:14:14] <cncuser> anyone more clou then me ? :)
[14:14:16] <jepler> cncuser: I think a 50Hz pulse would be pretty visible to a lot of people
[14:14:28] <cncuser> shure :)
[14:14:56] <cncuser> jepler: whats your favorite meal at the moment ?
[14:16:17] <cncuser> hmm, platform indipendent libraries
[14:16:23] <cncuser> with python
[14:16:44] <cncuser> $PYTHONHOME
[14:16:46] <jepler> cncuser: either thai yellow curry tofu or pizza with tomato sauce, pineapple, jalapeno pepper, and mozzarella cheese.
[14:17:13] <cncuser> thai yellow curry sounds like a good name for the rellease :)
[14:19:18] <jepler> but this LED thing is not from wall power, but from the regulated +5V supply
[14:19:42] <jepler> 18 strands of 2 series LEDs, with one resistor to set the current
[14:19:42] <cncuser> ah, i c
[14:20:12] <jepler> in case any of you care .. http://www.dansdata.com/caselight.htm
[14:21:20] <cncuser> hahaha
[14:21:27] <cncuser> in case if i dare ;)
[14:26:24] <cncuser> ok, somethings still missing with python, lets see
[14:30:17] <cncuser> jepler: is there a way to launch the axis gui without lanuching emc2 ?
[14:30:38] <cncuser> a complete start /stop circle takes 5 minutes in my qemu ;)
[14:33:47] <jepler> cncuser: I think that "cd (directory of configs); axis -ini (full path to config)" may do it
[14:34:25] <jepler> so something like: cd /..../configs/sim; axis -ini /.../configs/sim/sim.ini
[14:34:35] <jepler> giving a path to axis if necessary
[14:34:36] <cncuser> nice, thanks :)
[14:34:55] <jepler> I haven't done that with emc2, but I used to do something similar with emc1
[14:36:49] <alex_joni> jepler: thanks btw
[14:36:55] <jepler> alex_joni: for?
[14:37:16] <alex_joni> build system
[14:37:23] <alex_joni> saved me a lot of time yesterday
[14:37:33] <alex_joni> hacking at task, interp, canon, motion, etc
[14:37:48] <jepler> glad to hear it
[15:00:03] <alex_joni> omg, this is crazy: http://video.google.com/videoplay?docid=-760820435866344019
[15:04:05] <alex_joni> http://video.google.com/videoplay?docid=5393904704265757054
[15:08:30] <SWP_Away> SWP_Away is now known as SWPadnos
[15:08:37] <SWPadnos> I like that LED project, but I prefer Mariss' name for broken LEDs
[15:08:50] <SWPadnos> DEDs - "Darkness Emitting Diodes"
[15:13:08] <alex_joni> yeah
[15:20:34] <skunkworks> alex - I would say that the proccessor was shot off of the motherboard with a gun. fake. notice the hole through the motherboard.
[15:21:09] <SWPadnos> it was definitely faked - the board had probably been sawed beforehand, and hte processor blown off with a cap or the like
[15:24:36] <skunkworks> otherwise proccessors would have warnings. - "running this proccessor without heatsink could cause injory or death" ;)
[15:24:41] <skunkworks> injury
[15:24:53] <SWPadnos> they normally smoke a bit when they get hot
[15:25:13] <jepler> I noticed that the fan was not turning in the video
[15:25:30] <SWPadnos> and the force required to pull it out of the socket (if it's closed) is significant - the core would have popped out, but left most of the pins in the socket
[15:25:39] <skunkworks> right
[15:37:44] <alex_joni> yup, I agree
[15:38:04] <SWPadnos> but still a kinda funny video
[15:38:29] <skunkworks> do I see a mythbusters episode in the future?
[15:38:48] <SWPadnos> the magic 8-ball says - "unclear"
[15:39:02] <alex_joni> skunkworks: nah.. not yet
[15:39:15] <alex_joni> this is sweet :D http://www.eblogx.de/picdump/40/image.jpg
[15:40:13] <skunkworks> what was that artist name that had drawn those?
[15:40:23] <skunkworks> that is cool
[15:40:39] <skunkworks> I cant think of it now.
[15:41:02] <alex_joni> yeah, I was trying to remember aswell
[15:41:06] <alex_joni> esher something
[15:41:34] <alex_joni> escher?
[15:42:28] <alex_joni> http://www.worldofescher.com/store/P56.html
[15:43:27] <alex_joni> M. C. Escher
[15:44:37] <SWPadnos> http://www.worldofescher.com/gallery/A8.html
[15:45:27] <SWPadnos> there was a cool version of this one on the cover of BYTE magazine way back when:
[15:45:32] <SWPadnos> http://www.worldofescher.com/gallery/A13.html
[15:52:11] <jepler> back to the exploding CPU video, the screen shows a completely unlikely core voltage of 4.xxx, while I don't think any modern CPU uses a core voltage above 2.
[15:52:22] <alex_joni> overclocked ones do
[15:52:29] <alex_joni> but you need massive cooling for that
[15:52:34] <alex_joni> not a simple heatsink
[15:52:43] <SWPadnos> over 2 maybe, but definitely not 4
[15:52:58] <alex_joni> it's the only way to drive a 1Gig Duron up to 3.4Gigs (if I read that ok on the page)
[15:53:08] <SWPadnos> unless you used liquid nitrogen cooling or something ;)
[15:57:36] <jepler> using a stupid equation (http://www.heatsink-guide.com/content.php?content=maxtemp.shtml) for estimated power usage, I get 840W to overclock a 70W CPU by 3x and increase core voltage from 2V to 4V.
[15:57:52] <jepler> that just seems .. a bit too much
[16:04:17] <alex_joni> lots too much
[16:06:03] <jepler> http://emergent.unpy.net/files/sandbox/axis13.png
[16:06:32] <jepler> so what do you think? plot on a light background instead of dark?
[16:08:21] <cncuser> make it configurabele :) i find dark things with hue highlights to be more nice to my eyes
[16:08:40] <cncuser> console tyle ;)
[16:09:00] <cncuser> offcourse, a sunframebufferconsole is the oposit
[16:10:05] <jepler> actually it will be configurable in axis 1.3. I was playing with dark-on-light to show the new configurability.
[16:10:26] <cncuser> cool
[16:10:57] <jepler> I agree 100% about terminals and black backgrounds, but most other programs I use with light backgrounds.
[16:11:11] <cncuser> if i can configure i change
[16:11:20] <cncuser> like with beaver, anjuta, xchat
[16:11:47] <cncuser> but my eyes are rather sensitive too light anyways
[16:12:34] <jepler> of those, I only know what "xchat" is
[16:12:48] <jepler> (I use irssi, which runs in my nice black terminal)
[16:13:23] <jepler> I don't use a "development environment" either, I use vim in terminal.
[16:13:25] <jepler> so what's "beaver"?
[16:13:35] <cradek> jepler: I think with white background the program looks more "modern" but I don't like it as much.
[16:13:36] <cncuser> hehe, the days i kept my irc session running 24/7 in screens is over
[16:13:38] <jepler> Linux: 2.6.0 Released; "The beaver is out of detox"
[16:16:40] <cncuser> dear channelers and channeleresques
[16:17:16] <cncuser> i gonna put a new version of the livecd online as soon as i find some place to upload from :) get yout cdburners ready ;)
[16:19:05] <cncuser> anyone done a nice emc2 icon ?
[16:19:08] <SWPadnos> need a mirror?
[16:19:15] <cncuser> also axis icons i search for
[16:19:30] <cncuser> SWPadnos: shure.
[16:19:50] <SWPadnos> ok. if there's a place I can get the iso from, I can put it on the cncgear site
[16:20:12] <cncuser> SWPadnos: oh, cool, i thought about doing that too :)
[16:20:56] <cradek> yes there's an emc icon in cvs
[16:20:56] <cncuser> i need to merge the howto, paint some icons, anbd later in 3 or 4 hours i go to a internetconnection that doesnt get disconnected when uploading
[16:21:05] <cncuser> cradek: cool
[16:21:09] <cradek> I use it for the menu in the ubuntu pcakage
[16:21:18] <cradek> (let me know if you can't find it)
[16:21:28] <cncuser> emc2icon.png
[16:21:30] <cncuser> haha
[16:21:30] <cncuser> ok
[16:21:43] <cncuser> hehe
[16:21:48] <SWPadnos> axis also has an icon, at least there's a small window decoration on the window
[16:21:56] <SWPadnos> that made no sense
[16:21:56] <cncuser> ok
[16:22:02] <SWPadnos> I should have another pot of coffee
[16:22:03] <cncuser> its fenns design
[16:22:06] <cncuser> nice
[17:00:11] <Jymmm> GF's car got broken into last night.
[17:44:24] <CIA-8> 03jepler * 10emc2/scripts/emc.in: Allow extra commandline arguments to be passed on to the DISPLAY program. AXIS uses this to open an initial file (full path required)
[18:03:04] <dave-e> hello alex
[18:03:09] <alex_joni> hi dave
[18:03:51] <dave-e> anyone had any experience with the Mesa 5I20/7I33 combo
[18:04:23] <alex_joni> dave-e: not me..
[18:04:41] <alex_joni> how's the STG working? seen you were trying to get it running..
[18:04:48] <dave-e> channel 1 encoder just died on my stg
[18:04:54] <alex_joni> use another one
[18:04:59] <alex_joni> 8 should be plenty
[18:05:20] <alex_joni> with emc2 & HAL you can still use DAC1, IO, etc and ENC8 for example
[18:05:33] <dave-e> oh....that right I have hal...I keep forgetting
[18:05:47] <skunkworks> hal for the win
[18:05:58] <alex_joni> if you want I can walk you through that
[18:06:03] <alex_joni> it's fairly easy
[18:07:00] <dave-e> I can visualize it but doing it might be different
[18:07:30] <dave-e> basically assige encoder x to Y---
[18:08:02] <dave-e> in fact encoder 3 is not being used although the dac is..(spindle)
[18:08:33] <dave-e> so it is setp ....??
[18:10:07] <alex_joni> dave-e: you are acustomed to editing files, right?
[18:10:11] <dave-e> I need to download the latest changes to emc
[18:10:24] <dave-e> yes... usually with vi
[18:10:44] <alex_joni> ok, I won't hold that against you :D
[18:10:59] <dave-e> well as a friend of mine says....vi is vile
[18:11:05] <dave-e> but universal
[18:11:06] <alex_joni> look at configs/stg/stg_motion.hal
[18:11:18] <dave-e> ok
[18:12:13] <alex_joni> see:
[18:12:13] <alex_joni> # connect position feedback signals to encoders
[18:12:14] <alex_joni> linksp Xpos-fb <= stg.0.position
[18:12:42] <alex_joni> that links the position feedback of axis X to the position measured by the first encoder counter
[18:13:26] <dave-e> and so for Y linksp Ypos-fb<=stg.3.position to use channel 03 encoder
[18:13:41] <alex_joni> the stg.1.position is the one busted?
[18:13:50] <dave-e> yes
[18:13:55] <alex_joni> ok, then yes
[18:14:04] <dave-e> this is too easy
[18:14:12] <dave-e> but then that is the power of hal
[18:14:13] <alex_joni> there is another thing you need to do
[18:14:17] <dave-e> ok
[18:14:24] <alex_joni> read the next few lines.. you'll figure
[18:14:41] <alex_joni> figure it out, I mean
[18:14:48] <alex_joni> # connect position feedback signals to encoders
[18:14:48] <alex_joni> linksp Xpos-fb <= stg.0.position
[18:14:54] <alex_joni> er.. not that
[18:15:05] <alex_joni> # get feedback scaling from ini file
[18:15:05] <alex_joni> setp stg.0.position-scale [AXIS_0]INPUT_SCALE
[18:15:05] <alex_joni> setp stg.1.position-scale [AXIS_1]INPUT_SCALE
[18:15:06] <alex_joni> setp stg.2.position-scale [AXIS_2]INPUT_SCALE
[18:15:07] <alex_joni> # get feedback scaling from ini file
[18:15:09] <alex_joni> setp stg.0.position-scale [AXIS_0]INPUT_SCALE
[18:15:12] <alex_joni> setp stg.1.position-scale [AXIS_1]INPUT_SCALE
[18:15:14] <alex_joni> setp stg.2.position-scale [AXIS_2]INPUT_SCALE
[18:15:40] <alex_joni> * alex_joni is sorry.. too tired :(
[18:16:10] <dave-e> your time is about 1900?
[18:16:19] <alex_joni> dave-e: now, that you're not using stg.1.position anymore (but you are using stg.2.position), you need to do the same to the position-scale
[18:16:34] <alex_joni> dave-e: 20:17 (but I just got home from work) .. long day
[18:16:42] <dave-e> indeed
[18:17:00] <dave-e> so setp stg.3.pos.....................
[18:17:30] <alex_joni> yeap
[18:17:33] <alex_joni> you got it
[18:17:57] <dave-e> I think I can make it work....someday I might even understand this stuff.... ;-)
[18:18:09] <alex_joni> I bet you already do
[18:19:11] <dave-e> well not enough....tried to do a setp from hal_config....and it complained that I needed thread I didn't have
[18:20:44] <dave-e> I'm going to print this off and go try....thanks....have a good evening or what is left of it.
[18:20:50] <alex_joni> hrmm.. need a more exact paste to tell you what's wrong
[18:20:59] <alex_joni> dave-e: glad you can get it working ;)
[18:21:09] <dave-e> see ya later
[18:21:14] <alex_joni> I
[18:21:18] <alex_joni> I'll hang around
[18:22:08] <dave-e> bbl
[18:55:52] <cncuser> hi :)
[18:56:14] <cncuser> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?CncUser
[18:56:18] <cncuser> it works
[18:56:20] <cncuser> :)
[18:56:29] <cncuser> allmost copy and paste proof ;)
[18:57:46] <SWPadnos> does it actually require 2.5G of hard disk space?
[18:58:29] <cncuser> yes
[18:58:39] <SWPadnos> wow - that's a bunch
[18:58:43] <cncuser> well, hmmm
[18:58:58] <cncuser> maybe not
[18:59:03] <cncuser> ;)
[18:59:08] <cncuser> but 2.5gb is nothing
[18:59:21] <SWPadnos> it seemed like a lot for a distribution that fits into 50M :)
[18:59:26] <cncuser> hehe
[18:59:49] <SWPadnos> but you're right, it's about $1.50 worth of HD space these days
[19:00:09] <cncuser> you can put 3 on a dvd ;)
[19:00:30] <SWPadnos> I can put 120 on either of my larger hard drives
[19:00:48] <cncuser> :)
[19:00:56] <cncuser> ok, reboot time, cu in a second
[19:14:28] <Jymmm> GF's car got broken into last night.
[19:15:02] <Jymmm> $190 to replace the glass
[19:15:29] <jepler> Jymmm: blah
[19:15:37] <Jymmm> ?!
[19:16:00] <alex_joni> Jymmm: blah
[19:16:25] <alex_joni> ?!
[19:16:41] <jepler> !?
[19:17:10] <SWPadnos> !?!?
[19:20:03] <fenn> ...~
[19:20:44] <CIA-8> 03cradek 07stp_trap_blend * 10emc2/src/emc/kinematics/tp.c: report line numbers correctly
[19:28:08] <alex_joni> another one got scared away
[19:28:13] <alex_joni> Jymmm: wot's wrong?
[19:29:23] <Jymmm> alex_joni : Um ppl being asses for one.
[19:29:56] <dave-e> alex?
[19:30:17] <alex_joni> Jymmm: didn't mean that
[19:30:31] <alex_joni> I'm sorry bout that window..
[19:30:34] <alex_joni> dave-e: yes?
[19:31:11] <dave-e> most interesting.... shifted to channel 3 and put on a test encoder...and can only see the sign change
[19:31:34] <dave-e> ditto on channel,,,,, after I found a cable to add the next 4 channels
[19:31:41] <dave-e> channel 4
[19:31:48] <Jymmm> alex_joni thanks - the suck part is 1) they didn't even steal anything and 2) the feeling of being violated.
[19:32:10] <alex_joni> Jymmm: tell me about it.. although usually I got stuff stolen too
[19:32:21] <alex_joni> dave-e: only sign change?
[19:32:26] <dave-e> yes...
[19:32:28] <alex_joni> no numbers changing at all?
[19:32:31] <dave-e> no
[19:32:35] <cradek> Jymmm: I think random vandalism is almost worse than robbery. It's the senselessness that makes it so hard to understand.
[19:32:47] <alex_joni> can you look with halcmd at it?
[19:32:57] <dave-e> same thing
[19:32:59] <alex_joni> halcmd show pin stg.3.counts
[19:33:19] <alex_joni> I think it was counts
[19:33:22] <dave-e> oh... did it by looking at Y pos fb
[19:33:30] <Jymmm> alex_joni: She carrys a bag that LOOKS like a laptop (but isn't), they didn't go thru the car at all, just hit the trunk release and went looking for the bag, which wasn't a laptop. So it seem slike an inside job, someone that's seen her carry it.
[19:33:34] <alex_joni> look at the actual counts from the hardware
[19:33:53] <dave-e> ok
[19:33:55] <dave-e> brb
[19:37:25] <Jymmm> cradek: The sad part... cops dont even come out to take a report.
[19:42:52] <dave-e> ok raw counts work on 3 & 4 didn't check 1 but I'd bet it works also...
[19:43:05] <dave-e> so we have a disconnect someplace
[19:43:23] <alex_joni> yup.. most probably
[19:43:36] <dave-e> motion.hal ?
[19:43:52] <alex_joni> yup
[19:43:53] <SWPadnos> I wonder if there's a unit conversion issue somewhere - like (int) (counts * 0.001) = 0 (since it's an int)
[19:44:58] <alex_joni> no, it was working ok, befre moving to a different channel
[19:45:04] <dave-e> well I think I have all 3 axes set the same in the ini however I would't bet the farm on that
[19:45:55] <alex_joni> look at halcmd show pin stg.3
[19:46:02] <alex_joni> and paste all the info here if you can
[19:46:05] <dave-e> let me get the mo
[19:46:19] <alex_joni> if it's not too hard
[19:46:56] <alex_joni> dave-e: look at position too, not only counts
[19:47:25] <dave-e> that may take a bit....it is like everything else....200' one way to shop....get info..copy to keystick...bring up here ...download, et.
[19:47:59] <alex_joni> oh.. then you'll be my eyes ;)
[19:48:14] <alex_joni> look at stg.3.position, and see if that changes
[19:48:39] <alex_joni> if that doesn't change properly you don't have 'setp stg.3.position-scale [AXIS_1]INPUT_SCALE'
[19:49:30] <dave-e> I'm going to take a bit to offload the motion.hal file
[19:51:41] <cncuser> hi
[19:51:59] <alex_joni> hi again
[19:52:04] <cncuser> hin alex joni
[19:52:11] <alex_joni> und zur�ck
[19:52:20] <cncuser> im about to fit older inifiles to work with the new emc2
[19:52:36] <SWPadnos> making a conversion program?
[19:52:41] <gene> Hi all
[19:52:42] <cncuser> no
[19:52:47] <alex_joni> hi
[19:52:50] <cncuser> i hardly know where i should start :(
[19:52:55] <cncuser> emc barfs
[19:53:02] <alex_joni> cncuser: move it along, then start changing ;)
[19:53:08] <alex_joni> PERIOD -> BASE_PERIOD
[19:53:08] <cncuser> haha
[19:53:09] <gene> yup, won't build heer
[19:53:14] <gene> yup, won't build here
[19:53:25] <alex_joni> gene: what won't build?
[19:53:47] <gene> emc2, fresh cvs up -A
[19:53:55] <alex_joni> ???
[19:54:04] <alex_joni> it builds nicely for me.. what's wrong?
[19:54:31] <cncuser> Machine: EMC-HAL
[19:54:31] <cncuser> emc/iotask/ioControl.cc 846: can't load tool table.
[19:54:31] <cncuser> iniFind is depreciated
[19:54:31] <cncuser> HAL:66: ERROR: ini file variable '[AXIS_0]STEPGEN_MAXVEL' not found
[19:54:31] <cncuser> HAL config file /usr/local/etc/emc2/configs/CoolTool//core_stepper.hal failed.
[19:54:31] <cncuser> Shutting down and cleaning up EMC...
[19:54:37] <gene> I just posted on the emc list
[19:54:43] <SWPadnos> that's not a build problem
[19:54:53] <alex_joni> cncuser: you're missing STEPGEN_MAXVEL ;)
[19:54:57] <SWPadnos> oops - cross-read those lines
[19:55:02] <cncuser> yeah
[19:55:05] <cncuser> where to get it
[19:55:08] <cncuser> i dont have one
[19:55:08] <alex_joni> use 10-20% more then MAX_VELOCITY
[19:55:14] <alex_joni> write one ;)
[19:55:21] <alex_joni> look at stepper_mm.ini (from emc2)
[19:55:36] <cncuser> i hate it to do things i really dont have any coul off. like with emc configs
[19:55:57] <gene> I've found it can usuallu use 10x MAX_VEL here, same for MAX_ACCEL
[19:56:45] <gene> But thats just me giving it headroom
[19:56:51] <alex_joni> cncuser: if you wait a while the STEPGEN_ stuff might go away
[19:56:59] <alex_joni> gene: that's too much
[19:57:17] <gene> Now maybe, if I could get it to build
[19:57:18] <alex_joni> gene: what g+++++#
[19:57:21] <alex_joni> +
[19:57:25] <alex_joni> sorry..
[19:57:31] <alex_joni> what g++ are you using?
[19:57:52] <gene> Whatever is on a BDI-4.30 install?
[19:58:25] <cradek> 2.95
[19:58:26] <alex_joni> ++++++
[19:58:31] <alex_joni> aaargh.. sorry
[19:58:35] <gene> g++ (GCC) 3.3.5 (Debian 1:3.3.5-13)
[19:59:05] <cradek> gene: I answered your mail on the list already.
[19:59:36] <gene> It hasn't arrived here yet
[19:59:57] <cradek> gene: I've helped you with this problem before. If you make local changes to your cvs tree, you NEED to watch the "cvs up" output for merge errors and fix them.
[20:00:38] <gene> I haven't made any changes since it worked just fine last week, to anythng but the .ini file
[20:00:42] <cradek> oops, looks like jmk answered too, but he didn't think of a merge error.
[20:00:51] <cncuser> invalid inifile value for [TRAJ] LINEAR_UNITS: MM
[20:00:51] <cncuser> invalid inifile value for [TRAJ] ANGULAR_UNITS: DEG
[20:00:51] <cncuser> Traceback (most recent call last):
[20:00:51] <cncuser> File "/usr/local/bin/axis", line 1636, in ?
[20:00:51] <cncuser> lu = float(inifile.find("TRAJ", "LINEAR_UNITS"))
[20:00:51] <cncuser> ValueError: invalid literal for float(): MM
[20:00:53] <cncuser> Shutting down and cleaning up EMC...
[20:00:58] <cncuser> damit
[20:01:07] <cradek> gene: you applied a patch that jepler sent you, to enable M codes
[20:01:29] <cradek> gene: as soon as he puts that same fix in cvs, you can get a merge error.
[20:01:35] <cncuser> ok, here is something wron in my config...mm
[20:01:48] <gene> Mmm, yes I think I did, trivial patch really.
[20:02:29] <alex_joni> gene: cvs up -dPC ;)
[20:02:44] <alex_joni> the C in the end gets all the stuff from CVS, not from local files
[20:02:49] <gene> I just re-ran the cvs up -A, and don't see any squawks?
[20:03:03] <cradek> gene: you would have seen a C line (conflict)
[20:03:05] <gene> Oh, running
[20:04:27] <cncuser> emc/iotask/ioControl.cc 846: can't load tool table.
[20:04:30] <cncuser> ok
[20:04:38] <cncuser> got it all fixed, whats with the tooltable thing ?
[20:05:19] <cradek> // open tool table file
[20:05:19] <cradek> if (NULL == (fp = fopen(name, "r"))) {
[20:05:19] <cradek> // can't open file
[20:05:19] <cradek> return -1;
[20:05:19] <cradek> }
[20:05:38] <cradek> just what it says - it can't load the tool table
[20:05:49] <gene> Humm, looks like I'll need to restore my stepper_inch.ini file, it didn't like it, also the file I'd patched
[20:06:02] <cncuser> how does a tooltable look ?
[20:06:12] <cncuser> so that, if i see one, i can tell :)
[20:06:18] <cradek> gene: you can clean up a merge conflict manually. Your data is still in there surrounded by <<< >>>
[20:06:42] <cradek> gene: the idea is that you have to choose whether you want the changes from cvs, or your own changes
[20:07:03] <cradek> gene: and you should NOT modify a config that comes from cvs, for this reason, you should create a new config with a different name.
[20:07:34] <dave-e> alex.......problem solved...
[20:08:03] <dave-e> between vi and my typing INPUT_SCALE was set to 0
[20:08:07] <dave-e> for Y
[20:08:36] <dave-e> thank you very much .... it kept me from sending a good card back for repair.
[20:08:56] <dave-e> now to get cable down there so I can update
[20:09:04] <websys> dave-e - got your mail - was out of town - not sure what you mean by "getting fixture"
[20:09:23] <dave-e> oy...
[20:09:41] <gene> Looks as if it will bail out, same list of errors as before..
[20:10:00] <dave-e> being able to insert G55 -> 59.3 at the time of post
[20:10:22] <alex_joni> dave-e: glad to hear that
[20:10:30] <websys> They are passed thru when you use workzones
[20:10:35] <gene> Heres the script
[20:10:38] <gene> gene@shop:~/emc2$ cat updateEMC2
[20:10:39] <dave-e> ah
[20:10:40] <gene> #!/bin/bash
[20:10:42] <gene> cvs up -dPC
[20:10:45] <gene> cd src
[20:10:46] <gene> ./configure --enable-run-in-place
[20:10:47] <gene> make clean
[20:10:49] <gene> make
[20:10:51] <gene> sudo make setuid
[20:11:14] <dave-e> websys---I'm bringing up emc2 on mazak
[20:11:34] <websys> with toolchanger?
[20:11:42] <alex_joni> websys: sure ;)
[20:11:44] <dave-e> not yet
[20:11:56] <alex_joni> dave-e: but soon I hope
[20:12:10] <dave-e> my align spindle doesn't work right
[20:12:26] <dave-e> got to dig deep to fix it.
[20:12:46] <dave-e> will probably do that when the spindle goes in for rebuild
[20:13:41] <dave-e> I'm also working on vital...em2 for a BP sized mill
[20:13:46] <dave-e> emc2
[20:14:07] <gene> cradek: its acting a lot like the --enable-run-in-place isn't working
[20:14:11] <alex_joni> dave-e: soon we'll start asking you about emc2 :-P
[20:14:23] <dave-e> then I'm in trouble
[20:14:25] <alex_joni> gene: why don't you use debs?
[20:14:35] <cradek> alex_joni: he uses bdi
[20:14:39] <alex_joni> and?
[20:14:49] <cradek> alex_joni: there aren't emc2 debs for bdi, you know that
[20:14:59] <alex_joni> well.. he can use these I think
[20:15:11] <alex_joni> or I could make some debs for BDI..
[20:15:18] <cradek> no, that's a bad idea
[20:15:19] <dave-e> must say the halcmd show stg.n works well ... it was easy to verify that encoder 1, 3 & 4 were really working
[20:15:36] <cradek> he would need to change his kernel and rtai too.
[20:16:03] <cradek> gene: do you use emc1 - the one that came on bdi?
[20:16:25] <dave-e> paul made some noise a couple of weeks ago about a faster stepper driver for 4.38 .... has anyone gotten their hand on it. and tested?
[20:17:27] <alex_joni> nope, but I doubt it's faster than the one in emc2, maybe a bit
[20:18:01] <alex_joni> dave-e: I usually do 'watch --interval=0 halcmd show pin stg.'
[20:18:38] <alex_joni> that's like the poor-mans user-interface
[20:18:42] <gene> no, Ive always run emc2 since I had to re-install after a drive got cold
[20:18:59] <dave-e> on phone ...sorry
[20:19:22] <cradek> gene: why not upgrade to ubuntu then, so you can use the packages? you could update with cron, or by hitting "ok" in the update manager
[20:19:48] <gene> I'm logged into it nopw, instrffts
[20:19:55] <gene> instructions?
[20:20:03] <cradek> ?
[20:20:31] <gene> Oh, you mean junk the bdi-4.30 install?
[20:20:37] <jepler> dave-e: I got a copy of "quickstep" from the developer (whose name escapes me at the moment). I didn't do anything but look at pulses on a scope, but the results looked promising.
[20:20:54] <cradek> gene: yes, if you don't need emc1
[20:21:04] <jepler> dave-e: it's only for emc1, it would take heroic efforts to adapt it to emc2, based on what I saw reading the source code
[20:21:16] <jepler> dave-e: The code is GPL so I can forward you a copy if you like.
[20:21:34] <gene> Ok, but my ubuntu is old I exepct, url for a fresh .siio?
[20:21:34] <jepler> (I should say, it's only for bdi4emc1)
[20:22:05] <gene> fresh .iso :)
[20:22:17] <jepler> gene: cradek's packages work on ubuntu 5.10
[20:22:21] <jepler> I think it's called "breezy badger"
[20:24:02] <alex_joni> jepler: that's right
[20:24:17] <gene> yup found it, but np torrents for cd's, only dvd's
[20:24:43] <alex_joni> gene: you can also call for CD's
[20:24:51] <alex_joni> they send them to you
[20:24:58] <cradek> there is a CD torrent, I used it
[20:25:04] <gene> yup found it, but no torrents for cd's, only dvd's, tryin to eat a turkey sandwitch and type doesn't work all that well :)
[20:25:12] <cradek> there is a CD torrent, I used it
[20:25:28] <alex_joni> there is a CD torrent, cradek used it :-)
[20:25:30] <gene> let me look at the torrent linhk
[20:25:37] <jepler> gene: just use any of the download sites here .. http://www.ubuntu.com/download
[20:25:58] <cradek> http://ubuntu.cs.utah.edu/releases/5.10/ubuntu-5.10-install-i386.iso.torrent
[20:26:22] <cradek> ubuntu.com -> downloads -> any mirror site -> .torrent
[20:26:33] <jepler> oh yeah .. in the ftp directories there are torrents, but you can't tell because the extension is hidden in the ..>
[20:27:46] <jepler> mouse over them, luke
[20:28:16] <cradek> I must have been lucky because the mirror I chose showed the full filenames
[20:28:24] <jepler> aha
[20:29:08] <SWPadnos> jepler, can you see how the "quickstep" code differs from most any other step generation code?
[20:29:35] <SWPadnos> I think most of the time is spent in the RT task switch, not in the step generation code itself
[20:30:33] <jepler> SWPadnos: He installs a handler for the timer interrupt, which first outb's to the parallel port, then does the calculation of what the parallel port should be set to next time, and then returns (N-1)/N times, or jumps to rtai's entry point 1/N times.
[20:30:44] <SWPadnos> ok
[20:32:12] <jepler> SWPadnos: so there's a layer closer to the hardware than rtai which does the stepper pulses
[20:32:40] <SWPadnos> yep. that's the method I first used to do RT stuff on Linux (in the 2.0-2.2 era)
[20:34:36] <gene> Ok, got the torrent started for the install version
[20:38:27] <dave-e> jepler.....thanks for the offer of the code....I really don't need it...was just curious if anyone had tested...maybe even on real hardware. :-)
[20:39:00] <dave-e> gotta run....and concentrate on getting cable to the shop...
[20:39:20] <alex_joni> I tried running a papershredder with it.. that worked :)
[20:39:42] <SWPadnos> actually, that code should be OK with emc2, but getting the module load order right is a pain
[20:39:58] <SWPadnos> and you also have to make other modifications to the kernel
[20:40:48] <alex_joni> right, and I'm not sure if it's lots better ..
[20:40:50] <alex_joni> it could be..
[20:41:06] <SWPadnos> it should be - the work is done before checking for a task switch
[20:41:50] <SWPadnos> and you can eliminate the ultra-fast RT thread
[20:42:17] <SWPadnos> hmmm - but it won't work with HAL, since the output would be only for the parport
[20:42:38] <SWPadnos> it wouldn't be good to do the various pin-combining stuff at that level
[20:42:43] <alex_joni> yup.. so you get some performance increase, but lots of flexibility chumped off
[20:43:02] <SWPadnos> and the routine would need to know exactly where the step pulses should go, so other drivers would be out
[20:43:08] <SWPadnos> yes, speed vs flexibility
[20:43:21] <gene> A paper shredder? Aww, come on Alex, them are a $20 at wallies
[20:43:36] <alex_joni> gene: I was joking :/
[20:43:41] <SWPadnos> modern processors should be a lot better, but the AMD Elan SC520 I did this technique on had terrible jitter - in the 50 uS range
[20:43:49] <gene> And I of course had to nyte
[20:43:54] <gene> And I of course had to byte
[20:44:07] <alex_joni> the stuff was going into the shredder, not actually driving it :D
[20:44:21] <alex_joni> probably a bad joke :/
[20:44:23] <SWPadnos> that was for a 133MHz CPU (atually, a DX/4-133, so it had 33MHz I/O)
[20:44:41] <alex_joni> SWPadnos: ouch ;)
[20:44:47] <SWPadnos> yep
[20:44:53] <gene> thats more like it, now if we could just find one that would shred these damned credit card offers with the trial card in them
[20:44:58] <SWPadnos> it wasn't even suitable for a 1ms PID controller
[20:45:12] <alex_joni> gene: my shredder at work easily does credit cards
[20:45:27] <gene> Oh? Mine won't touch them things
[20:45:33] <cradek> my shredder does credit cards, but not CD ROMs
[20:45:51] <SWPadnos> my shredder does aluminum sheet
[20:45:52] <cradek> there were two models: I had to choose between cross-cut and CD ROM capable
[20:45:52] <alex_joni> it's one of those that doesn't produce long paper lines, but short ones
[20:46:00] <SWPadnos> it's a pair of Fiskars shears
[20:46:00] <cradek> yeah cross-cut
[20:46:04] <alex_joni> yeah, cross-cut
[20:46:07] <cradek> yeah
[20:46:09] <cradek> cross-cut
[20:46:09] <gene> A pocket knife across them a couple of times fixes them up real good...
[20:46:29] <alex_joni> who would want to shredd their pocket knife?
[20:46:37] <gene> I sense a pissin contest here... :)
[20:46:39] <SWPadnos> me me
[20:46:49] <alex_joni> cradek: I love wrecking cd-rom's by hand
[20:46:55] <SWPadnos> I'll shred Gene's pocket knife
[20:47:05] <alex_joni> with his shredder even
[20:47:11] <gene> Heavens, its a case!
[20:47:18] <SWPadnos> a nut case
[20:47:24] <gene> who me?
[20:47:39] <alex_joni> just the regular
[20:47:50] <SWPadnos> what - did I say something?
[20:48:03] <gene> mumble mumble, I guess not
[20:48:25] <SWPadnos> anyway
[20:48:36] <gene> seriously what happend to the cvs
[20:48:40] <gene> ??
[20:48:42] <SWPadnos> strange, I stil haven't received those emails to the emc list
[20:49:06] <alex_joni> gene: what's wrong?
[20:50:20] <gene> Got your message Chris, but whats wrong with: make && sudo make setuid?
[20:51:10] <gene> Its the make that bails out with all those can't find this and that error
[20:51:39] <alex_joni> gene: no it's not
[20:51:48] <alex_joni> try issuing just 'make', and see where it's barfing
[20:52:52] <SWPadnos> hmm - gonna try restarting mozilla. brb
[20:52:59] <gene> The && guarantees sequential operation
[20:53:55] <gene> make =:
[20:53:59] <gene> gene@shop:~/emc2$ cd src
[20:54:01] <gene> gene@shop:~/emc2/src$ make
[20:54:03] <gene> Compiling emc/rs274ngc/interp_convert.cc -fPIC
[20:54:05] <gene> emc/rs274ngc/interp_convert.cc: In member function `int
[20:54:06] <gene> Interp::convert_m(block*, setup*)':
[20:54:08] <gene> emc/rs274ngc/interp_convert.cc:1724: error: parse error before `<<' token
[20:54:09] <gene> emc/rs274ngc/interp_convert.cc:1733: error: ISO C++ forbids comparison between
[20:54:11] <gene> pointer and integer
[20:54:13] <gene> emc/rs274ngc/interp_convert.cc:1733: error: ISO C++ forbids incrementing a
[20:54:15] <gene> pointer of type `char*(*)(const char*, int) throw ()'
[20:54:17] <gene> emc/rs274ngc/interp_convert.cc:1733: error: non-lvalue in increment
[20:54:19] <gene> emc/rs274ngc/interp_convert.cc:1733: error: parse error before `)' token
[20:54:21] <gene> emc/rs274ngc/interp_convert.cc:1733: warning: statement with no effect
[20:54:22] <gene> emc/rs274ngc/interp_convert.cc: At global scope:
[20:54:24] <gene> emc/rs274ngc/interp_convert.cc:1740: error: `_setup' was not declared in this
[20:54:26] <gene> etc for about 3 screens full
[21:02:49] <alex_joni> gene: as cradek already said, you have a file (interp_convert.cc) which you locally modified
[21:03:09] <alex_joni> now that that exact modification is in CVS too, the cvs up procedure doesn't know how to merge them
[21:03:27] <gene> Yes, I found some merge trash in there that I just killed, it might build noiw
[21:03:28] <alex_joni> and you end up with some non-compileable files (interp_convert.cc in your case)
[21:03:40] <alex_joni> just delete the file, and let it come from CVS again
[21:13:08] <gene> Before I killed that file it did build, but complained about 3D_Chips.ncg, "bad character" after moving to 0,0,0
[21:13:27] <alex_joni> probably the same thing
[21:13:46] <alex_joni> you modified 3D_Chips.ngc locally, now you got merging problems
[21:14:29] <gene> And now the cvs up -dPC is hung on tcl/scripts for about 4 minutes now. And that file was moved aside and replaced by cvs
[21:14:35] <alex_joni> look for something like : '<<<<' in it
[21:17:33] <gene> I'm about to blow it all away and restart. Its still hung on tcl/scripts
[21:17:56] <alex_joni> gene: use 'cvs up -dPC'
[21:17:57] <cradek> gene: http://timeguy.com/cradek/emc/ubuntu
[21:18:02] <alex_joni> or that ;)
[21:18:57] <gene> I did use vcs up -dPC, and the torrent is running
[21:19:09] <gene> I did use cvs up -dPC, and the torrent is running
[21:19:42] <alex_joni> gene: the dPC is too late now, because the files are already merged
[21:22:21] <gene> Any fix?
[21:22:31] <cradek> remove the affected files
[21:23:24] <jepler> Is there a way to configure cvs to not change files that are locally modified (don't merge)?
[21:26:30] <gene> That would be nice... :)
[21:26:41] <gene> But now a new problem:
[21:27:01] <gene> HAL:66: ERROR: ini file variable '[AXIS_0]STEPGEN_MAXVEL' not found
[21:27:03] <gene> HAL config file /home/gene/emc2/configs/stepper//../common/core_stepper.hal failed.
[21:27:40] <gene> I had copied my stepper_inch.ini out of the way and then restored it, obviously somethings changed
[21:29:21] <alex_joni> restored it how?
[21:30:03] <gene> I maybe copied the wrong file, might have it now...
[21:30:17] <jepler> You can create new directories under configs/
[21:30:24] <jepler> cvs won't change files in those directories
[21:31:05] <gene> Got it, running now
[21:31:45] <gene> But what command do I use in this update script that will preserve my configs, but update EVERYTHING ELSE?
[21:32:16] <alex_joni> gene: don't use stepper_inch.ini
[21:32:21] <cradek> like several people have told you now, make new configs instead of changing the distributed configs.
[21:32:23] <alex_joni> make a new dir configs/gene/
[21:32:33] <alex_joni> and use that for your own configs
[21:32:37] <alex_joni> config
[21:33:15] <gene> Just got a follwoing error from joint 2. Ok, I can do that. But how do I make it select that automatcily?
[21:34:09] <gene> Gotta run and play chauffer, later guys & thanks
[21:35:08] <jepler> don't hit anything
[21:38:39] <gene> back in 20 mins
[21:40:08] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[22:46:18] <alex_joni> night all
[23:55:47] <gene> Trip was ok, got sidetracked in the shop putting it back together after putting the bigger yables on it.
[23:55:58] <gene> Trip was ok, got sidetracked in the shop putting it back together after putting the bigger tables on it.
[23:57:23] <gene> For the latecomers, I've put the LittleMachineShop expansion tables on my micromill that emc2 is running.
[23:58:03] <gene> Looks like everybody went to dinner, bye
[23:58:12] <bill2or3> Mmmmill