#emc | Logs for 2005-11-12

Back
[00:00:21] <LawrenceG> 50oz-in cont torq, about 150 peak
[00:00:27] <SWPadnos> I was looking into the possibility of "converting" a DC motor - like the treadmill motors they have - for servo use
[00:01:15] <LawrenceG> I have a pile of treadmill motors here I bought with the same idea.... adding the encoder is a pain, but it can be done
[00:02:08] <les_w> I did some analysis of what oem treadmill motors would do with some of the chinese cnc router components cropping up
[00:02:08] <SWPadnos> I found a set of three Baldor 28 in-lb motors on eBay for about $325 (with 100l BEI encoders)
[00:02:20] <SWPadnos> 1000 line, that is
[00:02:24] <LawrenceG> the treadmill motors would be a good splindle motor for a desktop machine... mine are 6000rpm rated
[00:03:22] <LawrenceG> hi les.. a few cool ones after thevisit from the suits??
[00:03:43] <les_w> Hi Lawrence. I found the better treadmill motors had about double the inrtia of typical servos of the same torque....still that would give acceptable performance for many applications
[00:03:55] <les_w> and the price is right even bought new
[00:04:22] <les_w> yeah visit went ok. I am funded.
[00:04:24] <SWPadnos> this one looks OK for that: http://www.surpluscenter.com/item.asp?UID=2005111017205381&item=10-1802&catname=electric
[00:04:27] <LawrenceG> yea... I think I paid 4.95cdn for mine.... not muct to them,
[00:06:19] <les_w> yeah that is one of the better ones. Square frame....about 0.5 oz in sec^2 inertia
[00:06:40] <LawrenceG> the surplus center 3ph servos came sealed in original sanyo boxes
[00:06:43] <les_w> it would do about 0.25G on the right structure
[00:07:04] <les_w> not too bad
[00:07:19] <icee> hey all
[00:07:22] <icee> it's unidirectional, though, isn't it?
[00:07:24] <SWPadnos> hiya
[00:07:43] <les_w> Hi icee...not sure on that one.
[00:07:44] <icee> oh, DC. nevermind
[00:08:10] <LawrenceG> it does say ccw... hey mike
[00:08:16] <icee> so lawrenceg and I have made some progress, too
[00:08:18] <les_w> even dc motors can have brush angle and timing for unidirectional
[00:08:29] <les_w> let me check again
[00:08:30] <icee> I've been sick the past couple of days, but with luck I'll have my test hardware up tonight
[00:08:39] <les_w> oh neat
[00:09:01] <icee> LG's hardware works fine except we have a stuck data pin for some reason.
[00:09:12] <SWPadnos> what's the hardware?
[00:09:15] <LawrenceG> cool... I ordered some samples from maxim of the line receivers
[00:09:39] <icee> SWP: a simple dsPIC board to test parallel port communication with
[00:09:50] <icee> we're designing a driver/amplifier for the 3 phase ac servos
[00:10:14] <SWPadnos> ah - OK. is the stuck bit on the parport or the PIC board?
[00:10:25] <LawrenceG> I am going to mill another simple board with just drive related stuff on it
[00:10:32] <icee> we haven't really determined, which is why I am going to replicate his test
[00:10:37] <SWPadnos> heh :)
[00:10:39] <icee> at this point i kinda suspect the parallel port
[00:11:02] <icee> we can toggle the pin without the parallel port connected, so..
[00:11:36] <SWPadnos> man - those are expensive PICs
[00:11:49] <LawrenceG> 12 free as samples!
[00:11:58] <SWPadnos> well - that's cheap
[00:11:58] <icee> swp: yah.. but they have nice 3 phase*2 PWM with dead time generators and quadrature decoders
[00:12:08] <SWPadnos> which one are you using?
[00:12:16] <icee> the 3011
[00:12:20] <SWPadnos> I assumed the 40-dip package
[00:12:40] <LawrenceG> 30f3011, 30f4011 has double the memory
[00:13:01] <LawrenceG> yep
[00:13:05] <les_w> here's a reversible treadmill motor
[00:13:07] <les_w> http://www.surpluscenter.com/item.asp?UID=2005111017205381&item=10-1483&catname=electric
[00:13:37] <SWPadnos> starting development is a real PITA, but the Freescale 56800E family is excellent for this kind of thing
[00:13:52] <icee> swp: I've used TI tms320f28xx
[00:14:00] <SWPadnos> they have BLDC drive software examples, and a plug-in board for motor drive
[00:14:02] <LawrenceG> hey... it would be fun to reverse a treadmill
[00:14:09] <icee> a lot of people are going to be building these for the sanyo denki motors though, so a requirement is everything throughhole
[00:14:15] <SWPadnos> the 320 series if more for calculations, the 56800s are for motor control
[00:14:20] <les_w> that one is no where near 1.25 hp of course
[00:14:28] <SWPadnos> that would put the kibosh on a DSP then
[00:14:30] <icee> swp: check out 28x; it's targeted at control
[00:14:41] <SWPadnos> which company?
[00:14:51] <icee> swp: tms320f28x, like tms320f2811
[00:14:56] <SWPadnos> ah
[00:15:06] <icee> on board program/data memory, PWM generators, quadrature decoding * 2, etc
[00:15:18] <SWPadnos> ah. similar to the 56800
[00:15:21] <icee> they're very nice parts
[00:15:43] <SWPadnos> the 56F805 has dual 6-phase PWM with deadtime, dual 20channel quad decoders, dual serial, etc.
[00:15:47] <icee> but they're in a very unfriendly .5mm pitch package
[00:15:52] <SWPadnos> 2 channel that was
[00:16:11] <SWPadnos> yes - that is the problem. at least they're not BGAs
[00:16:27] <icee> i was able to toaster oven them with a 75% success rate
[00:16:39] <icee> too nerveracking though
[00:16:44] <SWPadnos> heh - that's about 24.99% too low for production work :)
[00:17:00] <SWPadnos> PITA to probe as well
[00:17:08] <icee> well, if you're doing production work, contracting it out is cheap. but at qty 1 getting someone else to do it is 'spensive
[00:17:18] <icee> cheaper for me to trash a part here and there myself
[00:17:37] <SWPadnos> try Advanced Assembly - they're good and pretty inexpensive
[00:17:47] <SWPadnos> http://www.protosmtassembly.com/
[00:18:02] <SWPadnos> go through their online quote form - you may be surprised
[00:18:13] <SWPadnos> though they're better at SMT than TH
[00:18:24] <les_w> http://www.protosmtassembly.com/
[00:18:29] <les_w> oops
[00:18:32] <les_w> haha
[00:18:37] <SWPadnos> they even throw in the passives for free - easier than inventorying them
[00:18:56] <SWPadnos> les_w, I take it you've used them before :)
[00:19:38] <icee> I've used screaming circuits.. the prices seem similar
[00:19:45] <icee> are they set up to x-ray inspect BGA?
[00:19:49] <SWPadnos> never heard of them - got a link?
[00:19:54] <les_w> no just a mis paste. I actually hand solder smt, but I get paid well for it. I would just as soon still get paid well, but havw someone else do it.
[00:19:54] <LawrenceG> I see surplus center has the 100w sanyo motors on ebay
[00:20:14] <SWPadnos> these guys are down the street from Advanced Circuits, so delivery of PCBs to them is fast
[00:20:44] <SWPadnos> I used them on a recent project- it was less expensive than doing it by hand (for my customer)
[00:20:56] <les_w> I have about had it with soldering resistors the size of pepper grains
[00:20:58] <SWPadnos> plus I had an extra 5-8 hours to do other things
[00:21:27] <les_w> what did it cost if I may ask?
[00:21:28] <SWPadnos> I don't think they have the inspection tools for BGA, but they will place them
[00:21:51] <icee> * icee nods
[00:21:54] <SWPadnos> I got 5 boards with parts on both sides, 56 parts/board, and 0402 spacing, for $300-ish
[00:21:57] <icee> i only contract out bga stuff, generally
[00:22:06] <SWPadnos> let me check that though
[00:22:37] <les_w> Oh that sounds fine. I would have (and do) charge several times that,
[00:22:45] <les_w> better bookmark that one
[00:22:53] <SWPadnos> me, too. I just charge my normal $75/hour to do assembly
[00:22:57] <SWPadnos> that keeps them in line :)
[00:24:34] <SWPadnos> OK - it was a little more than I thought (of course, I had 3-day turn)
[00:25:12] <SWPadnos> $422 total, including FedEx priority shipping, 44 parts, 8 or 10 of which had to be hand-placed
[00:25:52] <SWPadnos> two sides (bi-color 0402 LEDs on one side, several 8-resistor 0402 spacing R-nets on the other side)
[00:25:53] <les_w> hand soldering 603 and 404 resistors by hand is not my idea of a good time...no matter what I get paid for it
[00:26:13] <SWPadnos> 0402 is about as small as I like to go by hand
[00:26:21] <SWPadnos> I've never even tried an 0201
[00:26:28] <les_w> seems I can farm it out and still make 75% as much
[00:26:31] <les_w> good deal
[00:26:42] <SWPadnos> yeah - or just do other things, and reap the benefits of that :)
[00:27:20] <les_w> as long as logistics don't eat up the time. With me, I just go out and do it.
[00:27:45] <les_w> if it's one...I wouls still do it
[00:27:51] <les_w> would
[00:28:00] <SWPadnos> if you place an order to DigiKey, just have it drop shipped. You'll get assembled boards a few days later (and leave out the standard passives, because they give them to you)
[00:28:11] <les_w> but usually do ten or so of something.
[00:28:25] <SWPadnos> 1-3, in house. 5-10, subcontract
[00:28:33] <SWPadnos> unless it's a 3-part board or something
[00:28:35] <les_w> yeah.
[00:29:14] <les_w> I get small multiples with mechanical stuff too...so I justify a VMC added to my shop.
[00:29:29] <les_w> also, I just want one. the r0outer is busy.
[00:30:24] <les_w> (a small one)
[00:30:57] <SWPadnos> heh - anything to justify new equipment, that's my mantra
[00:31:15] <les_w> told suits I needed a couple weeks off, so I have to scramble to get machines in, mount the new spindle on the router, and upgrade electrics.
[00:31:35] <les_w> Doubt if I will get everything done
[00:32:21] <les_w> Time flies by.
[00:32:31] <les_w> it was august a few days ago.
[00:32:38] <SWPadnos> yeah - just a few
[00:35:07] <SWPadnos> hmmm - there are Adeos patches for 2.6.12 on x86_64 - cool
[01:13:29] <SWPadnos> hey Ray
[01:13:40] <rayh> Hi Steve.
[01:13:40] <les_w> hi ray!
[01:13:50] <rayh> And les.
[01:14:02] <rayh> How'd the suits meeting go.
[01:14:03] <SWPadnos> (rhymes with Fay Wray)
[01:14:14] <SWPadnos> but not as good looking
[01:14:26] <rayh> Ain't that the truth!
[01:14:56] <les_w> good ray. I have funding. really good funding. We talked about me setting up manufacturing if the damn thing works. They likre the idea.
[01:15:20] <les_w> but where?
[01:15:41] <rayh> Awesome. rayh raises a glass of his favorite to les's good fortune.
[01:15:49] <les_w> ty
[01:16:04] <SWPadnos> yes - milk and cookies for everyone :)
[01:16:09] <SWPadnos> (my favorite)
[01:16:12] <rayh> That's it.
[01:16:14] <les_w> I feel pretty good. Could all be gone in a year, but now I am happy
[01:16:19] <les_w> and stressed
[01:16:30] <rayh> I know that feeling.
[01:16:39] <SWPadnos> heh -same here
[01:16:52] <SWPadnos> it's all good, until it ends
[01:17:06] <les_w> yup
[01:17:17] <rayh> The old mortens salt box comes to mind. When it rains it pours.
[01:17:28] <SWPadnos> yeah. ain't that the truth
[01:17:45] <les_w> I only got a couple weeks off from them to upgrade the shop.
[01:17:57] <SWPadnos> I had almost no work last year, and now, I'm designing a digital camera array, possibly a retail kiosk, doing other embedded work, and software as well
[01:17:57] <les_w> Let the tool buying begin.
[01:18:16] <SWPadnos> I need an engineering umbrella :)
[01:18:19] <rayh> It's fest time at les's?
[01:18:53] <les_w> First year, 2001, I borrowed $12k from My ira to live. didn't make a cent.
[01:19:09] <les_w> This year, 250 or so gross.
[01:19:21] <SWPadnos> 250k, I hope :)
[01:19:28] <rayh> Not bad at all.
[01:19:41] <les_w> next year perhaps zero again. Who knows.
[01:19:46] <les_w> yeah. k.
[01:19:54] <SWPadnos> nah - this project should last another 2 months :)
[01:20:10] <rayh> Let me offer a lesson from one of the north's celebs, les.
[01:20:12] <SWPadnos> (remember how late in the year it is :) )
[01:20:24] <rayh> Go out on top.
[01:20:40] <les_w> Too bad when I finally make some real money...it's worthless. Need a wheelbarrow of it to just get groceries.
[01:20:44] <rayh> I'm thinking of Green Bay's quarterback.
[01:20:53] <les_w> haha
[01:21:01] <rayh> who didn't
[01:21:31] <SWPadnos> Brett Farvre?
[01:21:44] <rayh> Yep. You got it.
[01:22:07] <SWPadnos> funny. I wouldn't have known his name if my in-laws weren't from Wisconsin :)
[01:22:25] <SWPadnos> I thikn it's required to be a Packers nut there :)
[01:22:32] <rayh> Aawesome in his day.
[01:22:45] <rayh> Or keep your mouth shut.
[01:23:17] <SWPadnos> heh
[01:23:33] <SWPadnos> I think they're on to that one. if you don't cheer at the right time, you're suspect
[01:24:35] <rayh> My advantage is that all three of the neighbors watch and cheer together so they can't hear me.
[01:24:41] <rayh> Not cheering.
[01:25:08] <SWPadnos> yeah, and you can just wait until they're drunk enough to not notice as well
[01:26:18] <rayh> * rayh switches boxes to work on cl and HAL for a bit.
[01:26:25] <rayh> be watching here from time to time.
[01:26:38] <rayh> rayh is now known as rayh-away
[01:33:25] <k4ts> night
[01:35:29] <Jacky^> g night
[01:38:45] <les_w> holy shit! A B-2 stealth bomber just flew ove r the house at 1000 AGL!!! I am out in the middle of nowhere!
[01:39:27] <Jymmm> les_w: Did it say ATF on the underside?
[01:39:41] <les_w> outta
[01:39:53] <les_w> first time I ever saw one
[01:39:58] <les_w> it is so quiet
[01:40:05] <Jymmm> les_w: Better hide those stiles of yours!!!
[01:40:12] <les_w> yeah
[01:41:15] <les_w> well, this is a MOA.
[01:41:24] <Jymmm> ?
[01:41:31] <les_w> Yikes!!! asecond one just flew over!!
[01:41:35] <Jymmm> Moose On Alcohol?
[01:42:16] <Jymmm> better hit the S/A and see whats going on
[01:42:17] <les_w> military operating area. F-18s regularly attack the caves on tigher mountain at very low level and high speed.
[01:42:59] <les_w> subsonic though.
[01:44:25] <les_w> But why would a b2 fly at 5 times the hwight of the trees?
[01:44:35] <les_w> height
[01:44:43] <Jymmm> training? testing something new?
[01:44:45] <les_w> 10 times sorry
[01:45:36] <les_w> something is gonna get blown up soon. You should have seen the activity here before shock and awe.
[01:46:13] <Jymmm> heh
[01:50:00] <SWPadnos> oh great
[01:50:18] <Jymmm> les_w it's alieans!!!
[01:50:24] <SWPadnos> "Let's bomb the ANWR, so nobody cares about drilling"
[01:50:37] <les_w> a B-2 looks like it. So quiet!
[01:51:08] <les_w> What do those things cost? 20 billion or something?
[01:51:20] <SWPadnos> the rumor was that the engineers at Lockheed knew they were on to something when they kept finding bats all around the planes in the morning :)
[01:51:30] <Jymmm> $2999.95 at WalMart
[01:51:30] <SWPadnos> on the floor
[01:51:42] <SWPadnos> $3009,99 at Costco
[01:52:01] <SWPadnos> out of stock at Home Depot
[01:52:19] <les_w> Well Lockeed Martin Dobbins AFB is about ten minutes flying time from here
[01:52:26] <les_w> 100 miles
[01:52:29] <Jymmm> only carries the accesories at home depto
[01:52:34] <SWPadnos> in slow mode :)
[01:53:08] <les_w> yeah...they were slow.
[01:53:22] <les_w> In a light plane it is about 20 minutes.
[01:53:32] <SWPadnos> looks like they're around $1billion each
[01:53:43] <les_w> oh...chep ride.
[01:53:47] <les_w> cheap
[01:53:49] <SWPadnos> in an SR71, it's about 100 seconds
[01:53:54] <SWPadnos> (at speed)
[01:53:55] <les_w> right
[01:54:16] <SWPadnos> and about twice that in a flat-out F14 or FA/18
[01:55:11] <SWPadnos> that reminds me. Les (or Jymmm), have you read "The New Pearl Harbor", by David Ray Griffin?
[01:55:24] <Jymmm> * Jymmm doesnt read
[01:55:30] <SWPadnos> oh right, I forgot
[01:56:01] <Jymmm> I might if ym gf gets the new Dean Koontz, but that's about it
[01:56:10] <les_w> have not read it
[01:56:19] <les_w> but I am a pilot
[01:56:30] <Jymmm> I play a pilot on TV
[01:56:33] <les_w> (of cessnas and pipers mostly)
[01:56:36] <SWPadnos> very interesting read, especially for the scientifically non-challenged
[01:56:56] <Jymmm> Lear Jets
[01:57:13] <SWPadnos> (regarding the 9/11 events, in case anyone doesn't know)
[01:57:43] <les_w> good read huh?
[01:57:48] <SWPadnos> yes, very
[01:58:13] <SWPadnos> the second book (about the 9/11 report) should also be pretty darned interesting
[01:58:51] <les_w> I like to fly. As of late I have not flown much because it is expensive. I can fly a little now though.
[01:58:51] <SWPadnos> one interesting point to ponder: In no other case has a steel-frame high-rise building been destroyed by fire
[01:58:53] <SWPadnos> ever
[01:59:14] <les_w> hmm
[01:59:33] <SWPadnos> yeah - hmmm
[02:00:17] <K`zan> * K`zan spends far too much time in flight sims :-)
[02:00:23] <les_w> well a-36 structural kinda goes to hell above 800f...
[02:00:38] <les_w> I remember kzan
[02:00:57] <SWPadnos> 1300 F for the common high-rise steels, apparently
[02:01:17] <SWPadnos> but they're also designed to 5x the expected load, in general
[02:01:41] <les_w> ohhh. Yield at wahat...1000 psi?
[02:01:57] <les_w> that range anyway.
[02:02:04] <K`zan> SWPadnos: How many other high rise buildings have been used as targets using commercial airliners, just another thought there :).
[02:02:30] <SWPadnos> actually, the Brits (though they didn't crash any planes into them) did do burn tests on several high-rises
[02:02:44] <les_w> We got a lot of restrictions about the sears tower in chicago after 911
[02:02:46] <K`zan> Theory is a wonderful thing :).
[02:02:59] <SWPadnos> and a 50-ish tory building in Kuala Lumpur was burned out - no collapse
[02:03:17] <SWPadnos> story, that is
[02:03:32] <SWPadnos> some waping of structural members, but no failures
[02:03:34] <SWPadnos> warping
[02:04:02] <SWPadnos> (incidentally, one other well-known high rise has had an airplane bump into it, that I'm aware of :) )
[02:04:15] <les_w> not sure..i'm an awro, not a civil
[02:04:20] <les_w> aero
[02:04:22] <les_w> haha
[02:04:48] <SWPadnos> so you do smoke, then :)
[02:04:53] <K`zan> well, with X tons doing ?500? kts with ??? gal of jet fuel, I don't find it the least bit odd.
[02:05:24] <SWPadnos> you should - read up on it with an open but skeptical mind, and you'll be left wondering what went on
[02:05:45] <les_w> it's a good bit of kinetic energy and a lot of jet-a kzan
[02:05:46] <K`zan> I doubt it, conspiracy theories are a dime a dozen.
[02:05:52] <SWPadnos> you do realize that 3 buildings collapsed that day, right?
[02:06:01] <SWPadnos> (at th e WTC)
[02:06:40] <les_w> vmax on those planes is about 320 kts indicated, but I don't think they cared much
[02:06:59] <SWPadnos> (also note that the A-Q theory is a conspiracy theory, just a more widely accepted one :) )
[02:07:11] <K`zan> WHatever shoving the quadtant to the wall in a dive got them...\
[02:07:23] <Jymmm> SWPadnos: All I have to say about it ---> http://www.freedomunderground.org/memoryhole/pentagon244.swf
[02:07:26] <les_w> yeah.
[02:07:34] <SWPadnos> yeah - that's a great video
[02:07:45] <les_w> looking
[02:10:24] <SWPadnos> this is a bit different than the original, but very close
[02:13:02] <les_w> watching...the only strange thing to me is why they would target that building
[02:13:21] <SWPadnos> act of war
[02:13:42] <SWPadnos> (since it's an obviously military target)
[02:14:19] <les_w> there was one other plane...white house or capitol. so this should have been too right?
[02:14:32] <SWPadnos> who knows?
[02:14:58] <SWPadnos> interesting to note that a large passenger plane managed to get through the defenses of the most heavily protected structure on the face of the planet
[02:15:14] <SWPadnos> with its own SAM defense systems
[02:15:22] <SWPadnos> and radar, etc
[02:15:57] <Jymmm> SWPadnos: Who says it wasn't the USA that attacked it?
[02:16:09] <SWPadnos> the USA, of course
[02:16:16] <Jymmm> wrong
[02:16:16] <les_w> I busted ohare class B once. I got busy with other things. I turned off the transponder.
[02:16:20] <SWPadnos> (at least, that's what the gubmint is saying)
[02:16:28] <Jymmm> lol
[02:16:39] <SWPadnos> oops
[02:16:47] <SWPadnos> did they scramble for you?
[02:17:00] <les_w> It was just a few seconds
[02:17:10] <SWPadnos> ah - the limit is 2 minutes or some such
[02:17:23] <les_w> turning off the radar transponder helps a lot Oi think
[02:17:35] <SWPadnos> (far exceeded on 9/11, and no other day, but that's a different topic)
[02:18:15] <les_w> Other student at the time busted it...black cars were waiting for them when they landed
[02:18:22] <SWPadnos> heh
[02:18:23] <les_w> students
[02:18:34] <SWPadnos> they don't like it when you do stuff like that
[02:18:41] <SWPadnos> (the black car drivers)
[02:18:47] <les_w> they have to shut down
[02:19:03] <les_w> prob costs hundreds of thousands a minute
[02:19:25] <SWPadnos> which?
[02:19:42] <les_w> you see, The airport I learned to fly at was 8.5 NM DME from ohare.
[02:20:05] <SWPadnos> ah - the one on the water (in the flight sims :) )?
[02:20:22] <les_w> that was meigs...now gone
[02:20:30] <SWPadnos> yes - Meigs field
[02:20:42] <les_w> it was shaumburg
[02:20:49] <SWPadnos> ah - never heard of that one
[02:20:59] <les_w> just a little airport.
[02:21:10] <SWPadnos> I fly through O'Hare these days (at least, I ride through)
[02:21:53] <les_w> yeah, you have to go through ohare and hartsfield to get to heaven.
[02:22:11] <SWPadnos> nah - we have direct flights to O'Hare
[02:22:49] <les_w> I never landed there myself...24 hr reservation and $250 landing fee.
[02:22:57] <SWPadnos> urk - that sucks
[02:24:00] <SWPadnos> anyway - don't mean to clog up the channel. read the book if you get the chance
[02:25:40] <les_w> ok
[02:25:50] <rayh-away> and something like 170k approach minimum speed.
[02:26:10] <rayh-away> emc related question follows.
[02:26:38] <les_w> kzan...set Fs to ohare. take off 27L. fly 260 8.5 DME. that's where I learned to fly.
[02:26:48] <rayh-away> Perhaps some of you have seen the displays like IO_Show where each pin is listed.
[02:26:50] <SWPadnos> oooh - an emc-relates question :)
[02:26:58] <SWPadnos> s/s/d
[02:27:01] <SWPadnos> /
[02:27:21] <rayh-away> There are little round indicators that look a bit like led's.
[02:27:30] <SWPadnos> yep
[02:27:56] <rayh-away> In HAL the bin pins and signals are either "true" or "false"
[02:28:09] <rayh-away> What color is "true" and what "false"
[02:28:52] <rayh-away> this is a trick question?
[02:28:57] <SWPadnos> good question, but it's probably true = on = 1,and false = off = 0
[02:29:14] <rayh-away> Yes but I asked for color.
[02:29:25] <SWPadnos> yes, it is a trick question :)
[02:29:45] <Jymmm> True==non-black
[02:29:45] <rayh-away> black=off white=on?
[02:29:52] <SWPadnos> in IOShow, do they alternate between red and green, or are they "bright" and "dim" in the same color?
[02:30:05] <rayh-away> alternate red and green
[02:30:06] <SWPadnos> (or black, and some color)
[02:30:11] <Jymmm> rayh-away: use a stop light
[02:30:25] <SWPadnos> OK - most likely that green is true, and red is false
[02:30:29] <rayh-away> rayh-away is now known as rayh
[02:30:34] <Jymmm> green on, go, true. Red off, stop, false
[02:30:54] <Jymmm> yellow oh shit
[02:31:05] <SWPadnos> yellow means "go very fast"
[02:31:31] <rayh> I should build a little fist with a raised finger
[02:31:47] <SWPadnos> you could patent that "smilie"
[02:32:19] <rayh> 0 and 1 are pretty unambiuous
[02:32:30] <SWPadnos> are you looking at HAL (port) output sing IOShow?
[02:32:36] <SWPadnos> using
[02:32:53] <SWPadnos> or is this a HAL version of IOShow that uses signals?
[02:34:42] <rayh> phone brb
[02:45:00] <rayh> This is a Hal version.
[02:45:07] <rayh> signals or pins.
[02:45:16] <SWPadnos> ok - then true should be green
[02:45:25] <SWPadnos> you can test it though
[02:45:39] <SWPadnos> just set a signal to true, and look at with IOShow
[02:45:42] <rayh> I wish that I could see links, then I could show a complete path.
[02:46:02] <SWPadnos> yeah - a "net hilight" halcmd
[02:46:10] <rayh> on the old, +5 was red and 0 was green.
[02:46:25] <SWPadnos> I'd test it. there's a constant block, right?
[02:46:40] <rayh> parameters?
[02:46:46] <SWPadnos> hold on a sec
[02:47:13] <rayh> What I'm thinking is that I can list and draw in a whole set of pins or signals
[02:47:31] <rayh> then loop on each for it's current value.
[02:47:35] <SWPadnos> there's a halcmd to show the current value of a signal, right?
[02:48:02] <rayh> show gives more detail than I want to parse.
[02:48:19] <rayh> but I do need to sort by direction.
[02:48:25] <SWPadnos> sure, but you can use it once to see if true=red or true=green :)
[02:48:57] <rayh> Oh I see what you are saying.
[02:49:17] <SWPadnos> yup - blocks.c has a constant block
[02:49:35] <rayh> I think that I'll stray from IO_Show's way.
[02:49:38] <SWPadnos> though it's a number
[02:49:57] <SWPadnos> a float number
[02:49:59] <SWPadnos> bummer
[02:50:24] <rayh> I really like the "bright" and "dim" in the same color idea.
[02:50:29] <SWPadnos> like an LED :)
[02:50:47] <SWPadnos> it's just as easy as red or green, just use two shades of the same color for the fill
[02:50:49] <rayh> exactly.
[02:50:57] <SWPadnos> and you can use red for output and green for input if you want
[02:51:03] <SWPadnos> (or the opposite)
[02:51:13] <rayh> I was thinking along those lines.
[02:51:36] <SWPadnos> that's one thing that LabView does really well (sort of) - LEDs
[02:51:43] <SWPadnos> for only $2000-$4000
[02:51:59] <rayh> Don't thank me folks, send money.
[02:52:11] <rayh> that's Dumb as a Post
[02:52:23] <rayh> p.O. Box 145
[02:52:27] <SWPadnos> which - me sending money, or LabView
[02:52:29] <rayh> Crystal Falls, MI.
[02:52:44] <les_w> writing that down...
[02:53:01] <rayh> I'm certain that labview will have a few things over on my HAL watcher.
[02:53:17] <SWPadnos> a couple, but it's missing some other stuff
[02:54:27] <rayh> back to it
[02:54:30] <SWPadnos> k
[02:54:37] <rayh> rayh is now known as rayh-away
[02:54:46] <CIA-5> 03jmkasunich * 10emc2/src/rtapi/rtai_rtapi.c: revert a previous change that caused type mismatch warnings - RTAI expects to pass an int to each task, not a long
[02:55:34] <SWPadnos> Hiya, John
[02:55:44] <jmkasunich> hi
[02:56:30] <les_w> hey jmk
[02:56:42] <jmkasunich> howdy les
[02:57:36] <SWPadnos> hey jmk, do you know if there are any successes with RTAI / adeos on x86_64, SMP, or SMP x86_64?
[02:58:04] <jmkasunich> you asking if the RTAI folks have RTAI working, or if we have EMC working?
[02:58:15] <SWPadnos> well - RTAI being a precursor ...
[02:58:25] <SWPadnos> I know there are x86_64 Adeos patches
[02:58:39] <SWPadnos> I'm having a harder time tracking down SMP info
[02:59:06] <jmkasunich> RTAI has UP and SMP schedulers, I think SMP support has been in there for along time
[02:59:21] <SWPadnos> OK - that may be why there's no recent mention of it ("it just works")
[02:59:40] <jmkasunich> I actually have an old SMP box here (dual 600MHZ P3's), but never got around to loading a distro on it
[02:59:45] <SWPadnos> did you ever sort out the trouble installing BDI on a SCSI system?
[02:59:58] <SWPadnos> (other than replacind SCSI with IDE)
[02:59:58] <jmkasunich> that is the same system
[03:00:02] <SWPadnos> hmmm
[03:00:23] <jmkasunich> the problem was with the version of kudzu (I think) that was on BDI-4.20
[03:00:33] <SWPadnos> I've been trying to get BDI to install under VMWare (for development, understanding that it won't be RT)
[03:00:43] <jmkasunich> there was a short lived bug in kudzu that prevented it from working with SCSI
[03:01:06] <jmkasunich> it was fixed within a week, but Paul was unlucky enough to grab the busted version :-(
[03:01:18] <SWPadnos> hmmm - maybe BDI just doesn't like the virtual SCSI "hardware" the VMWare presents
[03:01:23] <jmkasunich> later BDI's should be fine, but I never got around to testing it
[03:01:38] <SWPadnos> ok. was there a difference with text vs graphical install?
[03:01:40] <jmkasunich> what is VMWare anyway
[03:01:53] <SWPadnos> virtual machine emulator
[03:02:12] <jmkasunich> I don't think so, the problem was in the hardware detection code, I suspect that is common to both install methods
[03:02:15] <SWPadnos> http://www.vmware.com/
[03:02:20] <Jymmm> jmkasunich: It allows you to run one OS from inside another
[03:02:31] <Jymmm> jmkasunich actually, one or more
[03:02:38] <SWPadnos> many
[03:02:47] <SWPadnos> depending on RAM and disk space
[03:02:53] <jmkasunich> sounds like the same thing that adeos does (at a different level)
[03:03:04] <jmkasunich> dunno if they'd play nice together
[03:03:19] <SWPadnos> sort of. it's interesting to see the virtual PC "boot" in a window (complete with BIOS setup)
[03:03:28] <SWPadnos> yeah - IU was curious about that
[03:03:31] <SWPadnos> I
[03:03:34] <jmkasunich> ewwww
[03:03:38] <jmkasunich> that's just wrong
[03:03:45] <SWPadnos> it's really cool when it works
[03:03:53] <SWPadnos> which it seems to do quite well, actually
[03:04:03] <SWPadnos> I have a windows 200 pro install in a window
[03:04:09] <SWPadnos> 2000 (though it seems like 200)
[03:04:41] <jmkasunich> how can bios setup work that way
[03:05:07] <jmkasunich> does it fake out all the hardware or what?
[03:05:11] <SWPadnos> I don't know all the details, but it is a VMWare BIOS
[03:05:14] <SWPadnos> yes it does
[03:05:25] <SWPadnos> takes over the PCI and other detection VIOS entries
[03:05:27] <SWPadnos> BIOS
[03:05:36] <jmkasunich> IOW, what happens if I go into the bios and say "disable the IDE controller", or "change the boot order"
[03:05:43] <SWPadnos> fakes a BusLogic or LSI Logic SCSI card, IDE, serial, USB, parallel, etc
[03:05:56] <SWPadnos> you can do that - the network is also virtualized
[03:06:04] <jmkasunich> oh, it fakes the hardware too?
[03:06:08] <SWPadnos> you can test netboot on a single machine (complete with internal NAT)
[03:06:11] <SWPadnos> yes
[03:06:15] <jmkasunich> so inb() and outb() are faked?
[03:06:37] <SWPadnos> at some level, yes. I'm not sure exactly what they do once the OS is installed
[03:06:45] <jmkasunich> thats just wrong ;-)
[03:06:46] <SWPadnos> the guest OS
[03:06:47] <SWPadnos> heh
[03:07:05] <SWPadnos> no good for RT, but great if I can't or don't want an RT kernel on my main machine
[03:07:19] <SWPadnos> (a nice shiny new dual opteron)
[03:07:31] <jmkasunich> oooohhhh, shiny!
[03:07:59] <SWPadnos> yeah - 4G of shiny RAM, a 7800GT video card, and dual 24" 1920x1200 LCDs
[03:08:10] <jmkasunich> you musta won the lottery
[03:08:10] <SWPadnos> (very shiny ;) )
[03:08:24] <SWPadnos> nope - just waited 5 years for a good development contract
[03:08:30] <SWPadnos> OK - it is a lottery, I guess
[03:08:50] <jmkasunich> lol
[03:09:02] <SWPadnos> cool motherboard: http://www.supermicro.com/Aplus/motherboard/Opteron/nForce/H8DCE.cfm
[03:09:25] <Jymmm> And of course they'll put it on sale next week, or come out with an even better one for the same price.
[03:09:45] <SWPadnos> of course. Dell had a big sale on the monitors 2 days after I bought mine
[03:10:05] <jmkasunich> did you get yours from Dell?
[03:10:09] <SWPadnos> the newer CPU stepping became available (and the same price as the old) about a week later
[03:10:17] <SWPadnos> no - eBay. new
[03:10:35] <SWPadnos> still a better deal than from Dell, becausetheir deal was only for one
[03:10:50] <jmkasunich> now you need a bigger desk
[03:11:02] <SWPadnos> heh - I need to find my desk :)
[03:11:09] <jmkasunich> I have that problem too
[03:11:10] <Jymmm> SWPadnos: If I see it next week for 1/2 the price do you want me to tell you?
[03:11:24] <SWPadnos> the two LCDs are replacing 3 19" CRTs anyway,so not much more space
[03:11:32] <SWPadnos> yes - one can never have too many :)
[03:11:34] <jmkasunich> probably less
[03:11:42] <jmkasunich> (less space)
[03:11:52] <SWPadnos> actually, it should be 1/2 price this week, since I got it a few weeks ago
[03:12:02] <SWPadnos> less total area, similar total width
[03:12:18] <Jymmm> I'd like a LCD, but too low resolution
[03:12:21] <jmkasunich> less depth, leaves more space on the desk
[03:12:23] <SWPadnos> more screen area, and slightly more total pixels
[03:12:26] <SWPadnos> yes
[03:12:42] <SWPadnos> it's pretty cool to have a 3840x1200 desktop
[03:12:46] <jmkasunich> when I got my LCD it made a big difference
[03:12:53] <SWPadnos> (gotta find some good hubble photos for backgrounds)
[03:13:02] <SWPadnos> yeah - these are prety spectacular
[03:13:03] <jmkasunich> ("only" 1600x1200, but still nice)
[03:13:06] <SWPadnos> heh
[03:13:37] <SWPadnos> (btw - if I had won the lottery, it would be two dual-core opterons)
[03:14:34] <SWPadnos> on a separate note, have you looked into RTNet since the discussion on the Geckodrive list (about the G-Rex)?
[03:14:48] <jmkasunich> very, very briefly
[03:14:53] <jmkasunich> my plate is too full
[03:15:10] <SWPadnos> heh - I know the feeling (though I still manage to waste^W spend time here :) )
[03:15:10] <jmkasunich> got a Rabbit dev kit and book because they were on sale, half-price
[03:15:30] <SWPadnos> yeah - funny. I found out about that the day after I bought my kit, with no book
[03:15:37] <jmkasunich> got the G101 when I got my gecko drives (Mariss was kind enough to donate it)
[03:15:45] <jmkasunich> but haven't done anything with it
[03:15:49] <SWPadnos> I may get another one at 1/2 price, just for spite
[03:15:58] <SWPadnos> yep - he sent me one as well
[03:16:00] <jmkasunich> I'm not sure that deal is still on
[03:16:10] <SWPadnos> I wrote a serial downloader for him
[03:16:38] <jmkasunich> if/when I do anything with that I'll have to talk to you
[03:16:50] <SWPadnos> that would be good for me as well
[03:16:58] <jmkasunich> right now I'm trying to get back up to speed on Jon's USC
[03:16:59] <SWPadnos> di dhe send a G101/102, or just the G101?
[03:17:15] <jmkasunich> just the 101
[03:17:25] <SWPadnos> ah - OK.
[03:17:36] <SWPadnos> the carrier is a nice thing
[03:18:39] <jmkasunich> yeah, but I felt a little guilty getting the free 101 since I knew I wouldn't do anything with it for a long time, if ever
[03:19:24] <jmkasunich> * jmkasunich opens up the box and looks at the G101
[03:19:34] <jmkasunich> tiny thing ain't it
[03:19:51] <SWPadnos> heh. I think he's happy getting them out there though (especially to people who may make it do something)
[03:19:59] <SWPadnos> it's pretty small for what it does though
[03:20:05] <SWPadnos> I just don't like the rabbit much
[03:20:24] <jmkasunich> I did my first coding on a Z80 ;-)
[03:20:37] <SWPadnos> yeah, but some things are better forgotten :)
[03:21:08] <SWPadnos> I may make an ARM board for the socket
[03:21:21] <SWPadnos> or possibly an AVR, but the ARM would be better
[03:21:40] <SWPadnos> I wish Mariss had connected more of the Rabbit pins to the FPGA
[03:21:56] <SWPadnos> only having a 16-byte address window is a limitation, IMO
[03:23:09] <jmkasunich> hmmm, no polarity markings or keying, you can plug the rabbit module into the 101 two ways
[03:23:20] <SWPadnos> yes you can
[03:23:36] <SWPadnos> there is a mark, but it doesn't jump out at you
[03:23:36] <jmkasunich> which one is right?
[03:23:58] <jmkasunich> ethernet hanging off the edge, or ethernet in the center of the 101?
[03:24:22] <SWPadnos> the pin 1 mark on the rabbit module goes on the same side as the pin 1 mark on the G101 (pin 1->)
[03:24:23] <jmkasunich> ah, I think I found the mark
[03:24:46] <jmkasunich> so ethernet hangs off the edge
[03:24:53] <SWPadnos> yep
[03:25:05] <SWPadnos> and no mounting holes on the Rabbit
[03:25:41] <SWPadnos> it's a pretty positive contact socket, but I don't particularly trust it in an industrial environment
[03:25:42] <jmkasunich> unless it's a real hostile shock/vibe environment, you won't need any
[03:25:47] <jmkasunich> that header is pretty robust
[03:26:03] <SWPadnos> use it a second or third time, and it won't be
[03:26:19] <SWPadnos> (the one on my unit is a little sloppy already)
[03:37:22] <rayh-away> rayh-away is now known as rayh
[03:37:29] <jmkasunich> hey ray ;-)
[03:37:29] <rayh> Evening John.
[03:37:53] <SWPadnos> I think Ray may get the answer to his question now :)
[03:37:55] <rayh> Na. My nose is bigrn that.
[03:38:06] <SWPadnos> heh
[03:38:20] <jmkasunich> what question?
[03:38:22] <jmkasunich> * jmkasunich hides
[03:38:32] <rayh> I
[03:38:51] <rayh> I'm looking into building a graphical halcmd display.
[03:38:51] <SWPadnos> we
[03:38:54] <SWPadnos> you
[03:38:59] <SWPadnos> (sorry)
[03:39:00] <rayh> them
[03:39:14] <jmkasunich> tell me more
[03:39:32] <rayh> and I was thinking of colors for the little virtual leds.
[03:40:03] <rayh> I kinda decided to abandon the red green of the IO series in emc.
[03:40:14] <jmkasunich> pink and chartruse
[03:40:29] <rayh> In favor of bright red dark red and bright green dark gren.
[03:40:33] <jmkasunich> (swishy leds)
[03:41:02] <jmkasunich> bright = on, dark = off, I like... what is the diff between red and green?
[03:41:12] <SWPadnos> 200 nm or so
[03:41:15] <rayh> In and out
[03:41:42] <rayh> I have a small problem with getting that with list though.
[03:41:47] <jmkasunich> how do you know what direction a HAL sig is going?
[03:41:59] <rayh> I suppose I can parse the show
[03:42:27] <SWPadnos> no - you have to know what the pins actually are (I/O, motion, PID, etc)
[03:42:39] <SWPadnos> signals aren't in or out, they're both, since they connect inputs to outputs
[03:42:51] <rayh> for bits you have the type
[03:42:59] <jmkasunich> right - "in" and "out" are defined by how you use them
[03:43:01] <SWPadnos> (so it depends on whether you're looking at pins or signals)
[03:43:15] <jmkasunich> might be "out" from the motion controller, "in" to the PID
[03:43:24] <rayh> I'd also like to list and show links.
[03:43:47] <rayh> So directionality is only good for pins
[03:44:05] <SWPadnos> yeah - a signal has to have an output and at least one input connected for it to be useful
[03:44:25] <SWPadnos> (else it's a no-connect)
[03:44:44] <jmkasunich> show sig is one way to get some info
[03:44:49] <rayh> "if a tree falls in the woods..."
[03:44:54] <SWPadnos> no shit
[03:45:00] <SWPadnos> (oh - that's a bear)
[03:45:06] <jmkasunich> save neta is another way
[03:45:17] <rayh> save neta?
[03:45:38] <jmkasunich> "save nets with arrows"
[03:45:43] <rayh> I see that our HAL doc is getting a bit dated with this stuff.
[03:45:56] <rayh> and what does it save it to?
[03:46:04] <jmkasunich> it generates a newsig command for each signal, and a linksp for each pin linked to that signal
[03:46:08] <SWPadnos> I wonder if something like Grapviz could be used for connection diagrams
[03:46:18] <SWPadnos> Graphviz
[03:46:22] <jmkasunich> just sends it to stdout
[03:46:29] <rayh> okay.
[03:46:34] <richo> morning all..
[03:46:38] <SWPadnos> http://www.graphviz.org/
[03:46:48] <jmkasunich> the idea is that if you feed that output back into halcmd, it will re-create the configuration
[03:46:48] <rayh> so it appears at the same place the show and list commands return
[03:46:55] <jmkasunich> yea
[03:47:06] <jmkasunich> just a differnet format
[03:47:14] <jmkasunich> show is designed to be read by humans
[03:47:23] <jmkasunich> save outputs valid halcmd commands
[03:47:53] <jmkasunich> you can also do save net (no trailing a), that omits the arrows
[03:48:12] <SWPadnos> are the arrows equals signs or minus signs?
[03:48:24] <jmkasunich> ==>, or <==
[03:48:36] <SWPadnos> hmmm - take a look at graphviz
[03:49:06] <SWPadnos> if you make an output format that lists all connections as "name" -> "othername";
[03:49:29] <SWPadnos> with a very little bit of header, and a } for a footer, you can make pretty connection graphs
[03:50:20] <SWPadnos> and you can label the signals as well, using a slightly different format
[03:50:32] <SWPadnos> (click on the images for larger images, then on the large images for the source)
[03:51:40] <jmkasunich> it automagically lays it out on the sheet?
[03:51:49] <SWPadnos> I think so
[03:51:56] <jmkasunich> impressive
[03:51:57] <icee> yah, graphviz/webdot is awesome
[03:52:01] <SWPadnos> maybe I'll work on that sometime
[03:52:09] <jmkasunich> not quite perfect for HAL, but close
[03:52:16] <icee> I use it to draw state machines all the time
[03:52:21] <SWPadnos> take a look at the gallery - there are a number of formats
[03:52:46] <jmkasunich> main prob - all edges aka signals go to the a block, but there is no distinction about "where" on the block
[03:52:47] <SWPadnos> it wouldn't be useful for a full hal connection diagram, but certainly for parts of it, it would be great
[03:52:59] <SWPadnos> there is an arrow, though
[03:53:17] <jmkasunich> if a block is a PID loop, you don't know which signal connects to feedback, which one connets to command, and which one connects to enable
[03:53:41] <SWPadnos> wouldn't there be a named signal for each of those?
[03:53:56] <jmkasunich> yes
[03:54:08] <jmkasunich> HAL has named signals connecting named pins on named blocks
[03:54:13] <SWPadnos> and exactly one (or zero) outputs connected to that signal
[03:54:39] <jmkasunich> graphviz could easily show named signals connecting named blocks, but not which pins they connect to
[03:54:47] <SWPadnos> ah - right
[03:54:56] <SWPadnos> well - it's open source, I think
[03:55:08] <jmkasunich> yeah
[03:55:29] <jmkasunich> I've been thinking about the graphical tool, and the hardest problem is the layout one
[03:55:45] <SWPadnos> it looks like there is some support for connecting to a specific part of a block - look at the datastruct diagram
[03:55:51] <jmkasunich> if a human draws the diagram, then the file can contain both layout and connectivity info
[03:56:09] <jmkasunich> but if you start with a text file (or edit it), only the connectivity info is present (or valid)
[03:56:25] <SWPadnos> In a record label, curly braces { } enclose lists, vertical bar | separates list items, and creates a port identifier for attaching edges
[03:56:30] <SWPadnos> Edges are also labeled with 'id' attributes. Though not demonstrated in this particular file, these attributes allow referencing multiple (parallel) edges between the same node pair
[03:56:42] <SWPadnos> from the datastruct description
[03:56:42] <jmkasunich> cool
[03:56:51] <jmkasunich> which one is that?
[03:57:02] <SWPadnos> in the gallery, the third one
[03:57:29] <jmkasunich> ahh, I was looking at the examples on the main page
[03:57:41] <SWPadnos> yep - I noticed the gallery link on one of those pages :)
[03:57:49] <SWPadnos> (though it's on the main page as well)
[03:59:08] <rayh> I see that when I show or list funct I get a bit different name than I would use to show or list only one funct.
[04:00:05] <rayh> an example is classicladder.0.refresh but show understands classicladder or classicladder.0.
[04:00:12] <jmkasunich> SWPadnos: very, very interesting, bookmarked
[04:00:26] <SWPadnos> yeah - I just found it, and I like it
[04:01:10] <jmkasunich> rayh: not sure I follow
[04:01:35] <rayh> I'm writing a routine that burrows down into the details of hal.
[04:01:46] <jmkasunich> ok
[04:01:50] <rayh> I'm starting with a tabular set of windows
[04:02:00] <rayh> each tab would be a function
[04:02:05] <jmkasunich> "show funct" shows all functions
[04:02:15] <jmkasunich> "show funct foo" shows all functions beginning with foo
[04:02:32] <rayh> Right but the name it returns is not the name that would be used in subsequent show or list commands.
[04:02:42] <jmkasunich> its not?
[04:02:56] <rayh> Not quite.
[04:03:39] <rayh> some are but some are not.
[04:03:52] <jmkasunich> I don't have a CL setup right now, running vanilla emc.ini (steppers)
[04:04:00] <jmkasunich> and it seems correct here
[04:04:10] <jmkasunich> can you give me a specific example?
[04:05:16] <jmkasunich> John@main:~/emcdev/emc2head$ bin/halcmd list funct par
[04:05:16] <jmkasunich> parport.0.read parport.0.write parport.read_all parport.write_all
[04:05:17] <jmkasunich> John@main:~/emcdev/emc2head$ bin/halcmd show funct parport.0.read
[04:05:17] <jmkasunich> Exported Functions:
[04:05:18] <jmkasunich> Owner CodeAddr Arg FP Users Name
[04:05:18] <jmkasunich> 06 E0C2F230 E0C6F508 NO 1 parport.0.read
[04:05:19] <jmkasunich> John@main:~/emcdev/emc2head$
[04:05:52] <jmkasunich> I asked for a list of all functs starting with "par", got the list, then asked it to show info about the first one
[04:07:00] <SWPadnos> list gives the full name, show will accept subsets of the name
[04:07:10] <SWPadnos> (starting subsets, that is)
[04:07:16] <jmkasunich> list and show both use the specified name as a filter
[04:07:30] <rayh> the box I'm running on is separated from this so have to type from mem.
[04:07:39] <jmkasunich> both iterate thru all functions, and display only those that match the filter
[04:07:58] <rayh> parport.0.write and parport.0.read are returned by show funct or list funct.
[04:08:27] <rayh> but using either in show pin parport.0.write returns an error
[04:08:49] <rayh> while show pin parport.0. returns the pin names
[04:09:02] <jmkasunich> you did "show funct" which gave you function names
[04:09:11] <jmkasunich> then you tried to pass a function name to show pin
[04:09:18] <jmkasunich> there is no pin of that name
[04:09:21] <rayh> Yes I did.
[04:10:02] <jmkasunich> funct, pins, params, signals and threads are five independent lists
[04:10:09] <rayh> I'm old but not .....
[04:10:35] <jmkasunich> maybe I'm .... then, what are you trying to do?
[04:10:39] <rayh> the command "show pin iocontrol" will show all of the pins
[04:10:52] <rayh> in the function iocontrol
[04:10:57] <jmkasunich> all of the pins who's names start with iocontrol
[04:10:58] <SWPadnos> it's like ls iocontrol*
[04:11:04] <jmkasunich> exactly
[04:11:14] <SWPadnos> there's an implied asterisk after any parameter
[04:11:21] <jmkasunich> it doesn't "know" they belong to iocontrol, it just looks at the names
[04:11:24] <rayh> I know that it is just a bit of a stretch of your parsing code but it works well.
[04:11:26] <SWPadnos> (for show, list, etc)
[04:11:51] <rayh> All I can say is try it you may like it.
[04:11:54] <jmkasunich> it works for any module that uses a consistent naming convention
[04:12:07] <rayh> I do if only I could use it consistently.
[04:12:36] <jmkasunich> you want the real information about what modules export what pins....
[04:12:59] <jmkasunich> it breaks down for things like the motion module
[04:13:20] <jmkasunich> "show comp" shows you the names of the loaded modules, one is called motmod
[04:13:23] <jmkasunich> that's the motion controller
[04:13:33] <SWPadnos> (of course, you have to know that)
[04:13:40] <rayh> Then IMO we need to fix our reporting routines so that we can consistently get this kind of info.
[04:13:49] <jmkasunich> but "show pin motmod" doesn't show you the pins the motmod exported
[04:14:00] <jmkasunich> because they start with "axis", not motmod
[04:14:54] <SWPadnos> perhaps a list of these kinds of issues would be a good discussion point on Sunday ...
[04:15:29] <jmkasunich> the internal HAL data structures do keep track of which module exported which pins
[04:15:45] <jmkasunich> so a halcmd command to get that info could probably be implemented
[04:16:17] <SWPadnos> yeah - it looks like a comp->comp_id check would do it
[04:16:39] <rayh> comp doesn't give a very useful list either.
[04:16:46] <SWPadnos> fptr-->owner_ptr->comp_id
[04:17:18] <jmkasunich> ray: it's usefull, just nor for what you want to use it for ;-)
[04:17:23] <jmkasunich> s/nor/not
[04:17:47] <rayh> In my perfect world, I'd issue a command that would return module names
[04:17:58] <jmkasunich> kernel module names?
[04:18:04] <rayh> from these I'd build a set of tabbed frames
[04:18:26] <rayh> Hal components or functions
[04:18:47] <jmkasunich> ok - sometimes those map 1-to-1 to modules, sometimes not
[04:18:54] <rayh> I know that it gets messy with block stuff
[04:19:00] <jmkasunich> for example, module "blocks" can have many different things
[04:19:13] <rayh> No reason for a separate frame for each
[04:19:39] <jmkasunich> so basically you want to list the pins (or params, or functs) exported by a component
[04:19:44] <rayh> Once I've got tabs, then I'd like to fill up that frame with a list of pins
[04:20:08] <rayh> List works well for this if I know what to list xxx
[04:20:28] <jmkasunich> it sounds like you want a different filter
[04:20:29] <rayh> Then on each pin I can get and find the value
[04:20:46] <jmkasunich> the existing filter is just a string compare on the name
[04:20:54] <jmkasunich> you want to filter on the component that it belongs to
[04:20:56] <rayh> I'm surprised at how close we come to this with list.
[04:21:17] <jmkasunich> the naming conventions used by most (but not all) components give you what you want
[04:21:35] <SWPadnos> it's a trivial addition to halcmd, but I'd be more concerned about any semantic changes to the halcmd "language"
[04:21:48] <rayh> I suppose that this naming convention is clearer in stuff like iocontrol and classicladder
[04:22:08] <jmkasunich> SWP: you hit it on the head, halcmd is a language
[04:22:14] <SWPadnos> mayne halcmd {show,list} owner blocks
[04:22:23] <SWPadnos> mayne halcmd {show,list} owner name
[04:22:29] <jmkasunich> today:
[04:22:29] <SWPadnos> (blocks was an example)
[04:22:58] <jmkasunich> halcmd {show,list} <type> <name>
[04:23:36] <jmkasunich> <type> is pin, param, comp, funct, thread, orall, if omitted, it assumes all
[04:23:36] <SWPadnos> maybe tomorrow halcmd {show, list} [selector] <type} <name>
[04:23:56] <SWPadnos> with selector an optional parameter in {owner, child, ...)
[04:23:58] <SWPadnos> }
[04:24:10] <SWPadnos> ...connectedto...
[04:24:29] <jmkasunich> I was thinking more of keeping <name> as a single field, but extend the syntax
[04:24:51] <jmkasunich> <compname>:<name>
[04:24:54] <SWPadnos> <[domain:]name>
[04:24:58] <jmkasunich> the : would be reserved
[04:24:58] <SWPadnos> exactly :)
[04:25:00] <jmkasunich> yes
[04:25:06] <rayh> again in my imagination I see a list command like that that allows you to drill down from the topmost HAL organization level to the finest detail
[04:25:16] <SWPadnos> that's what we're talking about :)
[04:25:22] <jmkasunich> you take it further, with other possibilities other than just owner
[04:25:55] <SWPadnos> yes, like group, owner, child, class, connection, etc.
[04:26:13] <rayh> Thanks, guys.
[04:26:29] <SWPadnos> the concept of a group might be used by something like blocks to list the blocks loaded (and.0, and.1, not.0, etc.)
[04:26:46] <jmkasunich> not anytime soon
[04:27:04] <SWPadnos> sure - good to identify possible shortfalls before wider use
[04:27:12] <SWPadnos> heh
[04:27:15] <jmkasunich> the underlying code doesn't understand that the individual things loaded by blocks are indeed individual things
[04:27:24] <SWPadnos> does blocks know?
[04:27:31] <jmkasunich> not really
[04:27:46] <SWPadnos> one more item in the pin struct - a group number
[04:27:59] <jmkasunich> that's one of my pet peeves about hal 1.0, everything gets exported at insmod time
[04:28:20] <jmkasunich> you can't load blocks.ko, then export a sum2, then later export a mux4
[04:28:20] <SWPadnos> yep, and you can only load one module, so you have to know what you want ahead of time :)
[04:28:31] <SWPadnos> one copy of each module, that is
[04:28:55] <jmkasunich> that's something I want to fix in hal 2.0 (which right now is a separate project called blocs, with absolutely no completion date)
[04:29:15] <SWPadnos> completion date - wwhat's that? ;)
[04:29:26] <SWPadnos> one thing that we discussed in the last day or so was the idea of allowing "non-signal" connections - ie, pin to pin
[04:29:28] <Jymmm> jmkasunich: So, when will that be done? *DUCKS*
[04:29:47] <jmkasunich> pin to pin without a signal? why?
[04:29:48] <SWPadnos> I have a no-completion date. It's 11/10/2005
[04:29:59] <SWPadnos> it reduces the config file size by two thirds
[04:30:14] <SWPadnos> right now, you need 3 hal commands to connect a bit
[04:30:25] <SWPadnos> newsig, linksp, linksp
[04:30:29] <jmkasunich> ah, instead of newsig/linksp/linksp you would have linkpp
[04:30:34] <SWPadnos> yep
[04:30:40] <SWPadnos> or list connect or the like
[04:30:46] <SWPadnos> just, not list
[04:30:57] <jmkasunich> no
[04:31:07] <SWPadnos> or linkpp - whatever
[04:31:29] <SWPadnos> I was wondering if there's a fundamental reason why that can't be done
[04:31:31] <jmkasunich> the underlying code always needs a signal
[04:31:44] <jmkasunich> I can go into more depth if you want
[04:31:49] <SWPadnos> OK
[04:32:00] <SWPadnos> I remember discussions about memory allocation and shmem
[04:32:03] <jmkasunich> i have no real problem with a linkpp that internally does newsig/linksp/linksp
[04:32:08] <jmkasunich> but it has to make up a name
[04:32:11] <SWPadnos> OK - that ws my second choice :)
[04:32:27] <jmkasunich> could be as simple as N0001, like netlisters do for un-named nets
[04:32:41] <jmkasunich> (signals are really nets, using EDA terms)
[04:32:43] <SWPadnos> is there a limitation on signal names being unique from pin names?
[04:33:01] <jmkasunich> no, sigs and pins are separate namespaces
[04:33:02] <SWPadnos> (iw, can I have a parport.0.pin-7.in signal?)
[04:33:09] <SWPadnos> ok - that's solved then
[04:33:23] <jmkasunich> just use the name of the first pin
[04:33:32] <SWPadnos> the name of the output pin
[04:33:33] <rayh> I did quite a bit of common naming with iocontrol
[04:33:35] <SWPadnos> ther ecan be only one
[04:34:09] <jmkasunich> the pin that drives the net is the output pin?
[04:34:15] <SWPadnos> sure
[04:34:31] <SWPadnos> you can only connect one output pin to a signal, right?
[04:34:46] <jmkasunich> I'd rather use the first one - that lets the user decide which one is used
[04:35:38] <jmkasunich> for instance, if I connect parport.0.pin-11-in to axis.0.limit-sw-in, I might prefer the signal to be called axis.0.limit-sw-in, rather than parport.0.pin-11-in
[04:35:48] <SWPadnos> true
[04:35:49] <jmkasunich> or I might prefer the opposite
[04:36:23] <SWPadnos> yeah, but if you're interested in that, you'd probably be the type to use newsig/linksp/linksp
[04:36:36] <jmkasunich> maybe, maybe not
[04:36:49] <SWPadnos> I'm thinking of the person that has a sherline, and they want to connect x.step to parport.0.pin-3
[04:37:14] <jmkasunich> linkpp x.step parport.0.pin-3-out
[04:37:15] <jmkasunich> done
[04:37:21] <SWPadnos> yep
[04:37:23] <jmkasunich> the signal will be called x.step
[04:37:40] <rayh> Such a think linkpp would ease cl usage quite a bit as well.
[04:37:44] <jmkasunich> if they want it called parport.0.pin-3-out, they can reverse the order
[04:37:54] <SWPadnos> sure - I can accept that
[04:37:57] <jmkasunich> if they want it called anything else, they can use newsig/linksp/linksp
[04:38:29] <jmkasunich> * jmkasunich writes up a feature request tracker for linkpp
[04:38:32] <SWPadnos> it's only in the case of someone connecting multiple input pins to one output where my concern lies, and they can use newsig explicitly
[04:38:53] <jmkasunich> they can also do linkpp, followed by linksp's
[04:39:11] <SWPadnos> ooh - feature request tracker. maybe someone should put in the "show things owned by this component" idea
[04:39:18] <jmkasunich> I can see linkpp almost eliminating newsigs from config files
[04:39:23] <SWPadnos> sure, but this is like carpooling
[04:39:39] <SWPadnos> the first passenger to get in the car saves half the gas
[04:39:45] <SWPadnos> the next only another 16%
[04:39:47] <jmkasunich> yeah - that one needs a little more detailed definition before it can be implemented , but a tracker is a good place for that
[04:40:16] <SWPadnos> so for a single pin-pin connection (which the bulk of them are), linkpp saves 2/3 of the commands
[04:40:34] <SWPadnos> for any other combination, it saves only one command (or possibly two)
[04:40:42] <jmkasunich> two
[04:40:43] <SWPadnos> two, I guess
[04:41:09] <SWPadnos> so connecting 15 pins together (including halscope etc) isn't a big savings, though it's still convenient
[04:41:18] <jmkasunich> also eliminates the common (for me anyway) error of forgetting the type on a newsig
[04:41:23] <SWPadnos> yep
[04:41:34] <SWPadnos> take everything from the (first / output) signal
[04:41:42] <SWPadnos> pin, sorry
[04:42:34] <rayh> This would certainly make the raw hal file easier to read and follow.
[04:42:53] <SWPadnos> yep - I was noticing the 37k or so of Mazak configs
[04:43:04] <SWPadnos> thought that's not an average beast
[04:43:08] <jmkasunich> lotta comments in there too
[04:43:14] <SWPadnos> yeo
[04:43:16] <SWPadnos> yep
[04:44:07] <jmkasunich> I wonder if the save command should use linkpp whenever possible?
[04:44:10] <jmkasunich> probably should
[04:44:23] <SWPadnos> only if the signal matches a named pin, I'd think
[04:44:24] <jmkasunich> "possible" = signal name matches on of the pins connected to it
[04:45:04] <SWPadnos> yes - problems could arise if someone names a signal the same as a non-connected pin though
[04:45:15] <SWPadnos> you'd have to use newsig then
[04:45:31] <jmkasunich> well the save does this today:
[04:45:43] <jmkasunich> for (each signal) do
[04:45:54] <jmkasunich> phone
[04:48:57] <rayh> see you later
[04:49:04] <SWPadnos> see ya
[05:13:27] <SWPadnos> gotta run, myself - see you later jmk
[05:13:31] <SWPadnos> SWPadnos is now known as SWP_Away
[05:13:34] <jmkasunich> bye
[06:10:44] <CIA-5> 03paul_c * 10emc2-auto/wiki/ (37 files in 25 dirs): "Auto update wiki from a cron job. Fri Nov 11 05:30:01 GMT 2005 "
[10:17:51] <anonimasu> my mill arrived today
[10:17:51] <anonimasu> :
[10:18:07] <anonimasu> ^_^
[10:18:08] <alex_joni> and?
[10:18:54] <anonimasu> it couldnt be in better condition
[10:19:02] <alex_joni> great
[10:22:07] <anonimasu> yeah :)
[10:22:11] <anonimasu> pics to come later
[10:23:57] <anonimasu> the machine is very solid..
[10:23:59] <anonimasu> :)
[10:26:04] <anonimasu> *longs until he can take it for a spin
[10:26:30] <anonimasu> under cnc control
[10:26:30] <anonimasu> ;)
[10:27:14] <anonimasu> seems like the only backslash is in the couplings for the handwheels
[10:27:20] <anonimasu> and not in the screws
[11:02:38] <fenn> * fenn mumbles something or other about cvs
[11:06:31] <fenn> what does this mean:
[11:06:33] <fenn> cvs commit: failed to create lock directory for `/cvsroot/emc/emc2/src/hal/utils' (/cvsroot/emc/emc2/src/hal/utils/#cvs.lock): Read-only file system
[11:15:49] <alex_joni> hrmmm.. smthg is fishy on the CVS
[11:22:10] <anonimasu> hm
[11:22:22] <anonimasu> seems like the speed of the network at work just went up by a multiple of 5
[13:15:58] <fenn> if halcmd is screwed up, how do i remove hal modules? i keep getting "operation not permitted" when i try rmmod
[13:17:08] <fenn> oops i wasnt root n/m, now it just sits there instead of giving me an error :)
[13:19:53] <fenn> arg.. it appears there's nothing i can do
[13:52:37] <alex_joni> hey rayh
[13:57:02] <rayh> Hi alex
[13:57:23] <rayh> was on another box working on cl.
[13:57:34] <fenn> hacking on hal is no fun.. if you screw up you have to reboot!
[13:57:42] <rayh> I see that jmk is looking at linkpp
[13:57:46] <alex_joni> fenn: and that's no fun?
[13:57:59] <fenn> * fenn is trying to implement linkpp
[13:58:10] <rayh> Ah. Nice.
[14:06:10] <les_w> morning
[14:07:06] <les_w> I was hunting some MIC-6 aluminum cast tooling plate to mount the new spindle.
[14:07:08] <SWP_Away> fenn: think macro: newsig (arg1, typeof(arg1)); linksp (arg1, arg1), linksp (arg2, arg1)
[14:07:17] <les_w> found this:
[14:07:22] <SWP_Away> SWP_Away is now known as SWPadnos
[14:07:23] <les_w> http://www.sandsmachine.com/alumweb.htm
[14:07:34] <les_w> just though I would pass it along.
[14:07:50] <les_w> good price...think I will order 100 lbs or so
[14:08:24] <SWPadnos> that's the special Alcoa blens, right?
[14:08:27] <SWPadnos> blend
[14:09:11] <les_w> yeah real stable. ground or milled finish too. very accurate.
[14:10:07] <les_w> I was about to order some for $5/ lb...
[14:11:11] <les_w> this is $1.50/lb in 100 lb...
[14:11:15] <les_w> cheap!
[14:11:17] <fenn> SWPadnos: takes a while to thrash out the little details when you aren't familiar with the source
[14:21:17] <rayh> Name that movie, "I never mind the little things."
[14:24:00] <fenn> * fenn strangles hal
[14:27:16] <anonimasu> hm
[14:28:16] <fenn> anyone know how to kill a runaway hal?
[14:29:55] <SWPadnos> Big Red Switch :)
[14:30:06] <rayh> I'm a HAL newbee
[14:30:10] <anonimasu> hm
[14:30:13] <anonimasu> use a hammer
[14:30:14] <SWPadnos> The Shining?
[14:30:50] <alex_joni_> fenn: bin/halcmd stop
[14:31:00] <alex_joni_> then reboot.. if you can't rmmod the module
[14:32:49] <rayh> La Femme Nikita.
[14:32:58] <SWPadnos> hmmm - you would want to be sure that the pin specified is unique, as well (lest you get slightly different configs from run to run)
[14:33:05] <SWPadnos> Ah - haven't seen that one
[14:33:27] <rayh> There was a Hollywood remake.
[14:33:44] <SWPadnos> Yep - nice cover :)
[14:33:46] <alex_joni_> alex_joni_ is now known as alex_joni
[14:34:09] <rayh> It is in a scene where the "cleaner" has just killed her partner.
[14:34:35] <SWPadnos> "oops - silly me" :)
[14:34:53] <rayh> gotta get on the road.
[14:35:15] <rayh> good luck fenn
[14:35:20] <fenn> you too
[14:35:22] <SWPadnos> have fun Ray
[14:35:42] <fenn> i dunno why it's locking up at all
[14:35:54] <alex_joni> fenn: what's happening?
[14:36:43] <fenn> i'm making a new signal with the type of the first pin
[14:36:55] <fenn> then halcmd doesn't accept any more commands and i can't kill hal
[14:37:14] <fenn> i think it's because i still have the mutex
[14:37:47] <fenn> so i guess it's stuck in do_newsig_cmd somewhere
[14:38:33] <alex_joni> can you paste the snip code in a /msg ?
[14:39:15] <SWPadnos> can you just call the existing functions (hal_signal_new and hal_lnik)
[14:39:44] <alex_joni> that's what he does
[14:40:11] <SWPadnos> OK - those take care opf the mutex themselves
[14:40:43] <alex_joni> fenn: you need to release the mutex before the call to do_newsig_cmd()
[14:40:58] <alex_joni> because there is only one mutex
[14:41:02] <alex_joni> and you already got it
[14:41:16] <alex_joni> and do_newsig_cmd() will try to get it by itself.. and will fail
[14:41:32] <alex_joni> because your function doesn't release it..
[14:41:37] <alex_joni> got it?
[14:41:39] <fenn> ok
[14:41:52] <fenn> so can i recompile it and make it give up the mutex so i dont have to reboot?
[14:42:30] <fenn> only one way to find out
[14:42:55] <anonimasu> fenn: bye bye
[14:42:56] <anonimasu> :D
[14:43:21] <SWPadnos> nope - the new program won't be able to get the mutex :)
[14:44:02] <SWPadnos> you might be able to force rmmod the HAL modules, but I suspect that's dangerous
[14:44:18] <Jacky^> damn
[14:44:28] <Jacky^> [ot] http://www.beppegrillo.it/english.php
[14:44:31] <alex_joni> SWPadnos: won't work ;)
[14:44:37] <alex_joni> trust me .. been there, done that
[14:44:43] <SWPadnos> dangerous, right ;)
[14:44:45] <Jacky^> ghghghg
[14:44:48] <alex_joni> nah.. doesn't work
[14:45:02] <fenn> rmmod just bleeds the cpu dry
[14:45:11] <SWPadnos> even if you have "force module unload" in your kernel config?
[14:45:48] <SWPadnos> oops
[14:45:56] <Jacky^> absurd
[14:48:24] <SWPadnos> That's a terrible photo. I wonder where it's from.
[14:48:44] <Jacky^> * Jacky^ nods
[14:48:52] <Jacky^> n Falluja the civilian population was killed with white phosphorus that is like polite napalm
[14:49:03] <SWPadnos> no, it isn't
[14:49:17] <SWPadnos> white phosphorous gets about as hot as the surface of the sun
[14:49:21] <Jacky^> thats absurd
[14:49:33] <SWPadnos> it melts through engine blocks and the like
[14:49:50] <SWPadnos> so somebody found a nasty corpse (or more than one), and put clothing on it (them)
[14:50:15] <SWPadnos> I'm assuming that the text is sarcastic, because it's wholly inaccurate
[14:50:27] <SWPadnos> (regarding white phosphorous)
[14:50:48] <Jacky^> looking at video ..
[14:52:48] <SWPadnos> downloading - slow going.
[14:52:59] <Jacky^> me too
[14:53:07] <SWPadnos> It looks (from the text) like this is a video being aired in the "arab world"
[14:53:27] <SWPadnos> so, like Fox news, I'd be suspicious of its accuracy
[15:00:57] <alex_joni> * alex_joni goes away
[15:04:34] <anonimasu> hm
[15:04:42] <anonimasu> have anyone had a look at using a jogwheel with emc?
[15:04:44] <lerman> Re linkpp: Is is really worth the effort? To save a few lines of code in a configuration file that is created once at system build/integration time and takes essentially no time to execute seems like a waste of effort to me.
[15:05:31] <SWPadnos> configuration is the first or second most difficult part of using EMC for new users
[15:05:40] <SWPadnos> the other being installation
[15:06:49] <lerman> I suspect that the new command will not make the conceptual effort smaller. In fact, it might make it harder because the signal 'magically' gets its name from the first pin.
[15:07:22] <SWPadnos> true. it depends on the person.
[15:07:45] <SWPadnos> but looking at a 20-line .hal file has got to be easier than looking at a 60-line file for most people
[15:08:32] <lerman> The person who understands the signal abstraction will do fine in either case. The person who has trouble configuring might find it harder this way.
[15:09:01] <lerman> Particularly if he needs some signals that go to more than two places.
[15:09:19] <SWPadnos> I suspect that most users think in terms of connections, not "netlists"
[15:09:40] <lerman> But, who am I to argue -- if someone is willing to do the work and I don't have to use it.
[15:09:50] <SWPadnos> John K was thinking like a schematic editor, but most people just want to connect the X asis step output to a parallel port pin
[15:09:56] <SWPadnos> heh - right :)
[15:10:48] <lerman> But as soon as they want to connect it to a third place, they might have trouble because they will screw up the signal name.
[15:11:15] <SWPadnos> could be.
[15:11:18] <lerman> Ah... is John K building a schematic editor?
[15:11:36] <SWPadnos> no, but he thinks in schematic terms
[15:11:48] <SWPadnos> ValarQ was making a schematic-like editor
[15:12:23] <SWPadnos> using tools to generate the config, it doesn't matter what the underlying grammar is (for the most part)
[15:12:54] <lerman> Well, John K suggest naming the signal for the first pin listed. I would name it for the OUTPUT signal (that's the driving signal).
[15:13:13] <SWPadnos> That was my suggestion as well, since there can only be one of them
[15:13:30] <SWPadnos> my original idea was just to have any output pin actually *be* a signal
[15:14:10] <lerman> Then linkpp could list as many pins as you want. It would in effect, be a netlist command.
[15:14:14] <SWPadnos> the signal layer (as an indirection layer in HAL) helps if you want to move a group of inputs to a different output
[15:14:52] <SWPadnos> actually, it could do that anyway, but I think the halcmd interpreter isn't set up for varargs (effectively)
[15:15:47] <SWPadnos> it's a pretty direct parser -> if (strcmp(tokens[0], "linksp") ....
[15:16:42] <SWPadnos> maximum of 20 tokens as well
[15:16:49] <lerman> When I build a system like hal, I also used a signal abstraction. It is easier for a user to view a signal as: X drive, Y drive, Z drive instead of pid-1-output, pid-2-output, pid-3-output. (Actually, I have no knowledge of HAL naming; the previous is similar to what I did in an earlier life.)
[15:18:04] <lerman> build->built
[15:18:25] <SWPadnos> yep
[15:18:46] <SWPadnos> the extra layer also helps when you want to change things for test purposes or the like
[15:19:18] <SWPadnos> there are some inefficiencies in HAL right now.
[15:19:22] <lerman> As in connecting the signal to a virtual oscilloscope.
[15:19:28] <SWPadnos> yep
[15:19:55] <SWPadnos> like the fact that each pin actually stores its entire name: pid.0.loop.gain (or whatever)
[15:20:07] <SWPadnos> there's no inheritance of naming
[15:20:48] <SWPadnos> it makes some things much easier (like halcmd), but seems "icky" to me as a programmer
[15:21:05] <lerman> Well, memory is cheap :-) -- I'm a parser person, so I tend to be more interested in language design and ease of use.
[15:21:33] <SWPadnos> memory (and processing speed) are cheap, but that doesn't make it less icky :)
[15:21:38] <lerman> Yes. 'icky' is a good word.
[15:21:56] <SWPadnos> you also end up with things like pin names that don't contaqin the component name
[15:22:46] <SWPadnos> jmk's example was motmod - it has names like axis.0.velocity, not motmod.0.axis.0.velocity
[15:23:11] <SWPadnos> so you can't find it with a text search for motmod (which is what "show comp" gives you for the component name)
[15:23:57] <lerman> The pin structure should have a pointer back to the component to which it is a pin and get the name (prefix) from there.
[15:24:36] <SWPadnos> it does, but the halcmd language has no provision to use it (yet)
[15:24:54] <SWPadnos> take a look back through your log, starting around 22:30 last night (EST)
[15:25:07] <SWPadnos> jmk, ray, and I were discussing it
[15:25:25] <SWPadnos> the idea is to change the current syntax:
[15:25:53] <SWPadnos> {show,list} <name> <type>
[15:26:26] <SWPadnos> to {show,list} <[component:]name> <type>
[15:26:50] <SWPadnos> or more generically <[name:]name> <type>
[15:27:52] <SWPadnos> sorry - I was wrong a few lines back - the pin doesn't get its prefix from the parent - I misread your statement
[15:30:36] <lerman> Just read the earlier stuff.
[15:31:30] <SWPadnos> thoughts?
[15:33:59] <lerman> The details of the discussion are beyond me -- I really don't know much about the present HAL user interface.
[15:34:52] <SWPadnos> me either - I was just looking at the source :)
[15:34:55] <lerman> But -- adding complexity to the parser so as to simplify the use would be my approach.
[15:35:36] <SWPadnos> yep. it's a real deficiency that you can't reliably find out absolutely everything about the state of HAL through programmatic means
[15:35:51] <lerman> On the other hand -- there is so much to do in the area of additional functionality, that cosmetic changes to hal would be low on my list.
[15:35:59] <SWPadnos> (ie, knowing that a pin of a given name belongs to a component of a given name, where the names don't share anything)
[15:36:29] <SWPadnos> e.g. threading?
[15:36:34] <lerman> That is something worth fixing. -- although presumably, you could look at the command file to extract that info.
[15:36:49] <SWPadnos> assuming that only one command file is loaded
[15:37:07] <SWPadnos> you can losd one command file, then another (from the command line)
[15:37:11] <SWPadnos> load
[15:37:12] <lerman> Threading isn't on my personal lists -- probably because I don't need it. But I would support those who need it.
[15:37:27] <SWPadnos> you can then issue commands manually, and then you can't figure out what's connected where
[15:37:50] <SWPadnos> I thikn it's the biggest "marketing" problem. also not too important for me as yet
[15:38:40] <lerman> I think that most people who look at mach and look at emc don't see that as the most important distinguishing feature.
[15:38:55] <SWPadnos> feature-wise, Iwas about to write
[15:39:05] <SWPadnos> though the UI configuration ranks right up there for some reason
[15:39:26] <lerman> OTOH, if you nead threading, emc rules itself out.
[15:39:51] <SWPadnos> (I meant configurable UI, but also, the fact that you can configure things in the UI is important)
[15:41:31] <lerman> Yup. UI is ranks up there -- but not that it is configurable. Conversational mode would be high on my list. Rectangular pockets, bolt circles, boring, surfacing, would be nice to have without writing a program.
[15:41:43] <SWPadnos> yes
[15:42:20] <SWPadnos> I had thought about that as well.
[15:42:35] <SWPadnos> I was thinking of pendant control with an encoder wheel
[15:42:58] <SWPadnos> and then doing things like allowing you to change the angle of cut (rather than just linking to a single axis)
[15:43:52] <SWPadnos> and then saving a list of the moves you've done (collapsed where possible)
[15:44:40] <lerman> I'm starting to look at the interface from the gui to the interpreter/runtime so that I can start building my own gui. Now that I have subroutines, building this stuff should be simple. Each of those things (in the mach extensions -- I think they call them wizards) could be done with a single subroutine. So the gui would just display something that let you change the arguments to a subroutine call
[15:44:41] <lerman> and invoke the subroutine.
[15:45:02] <lerman> A pendent with encoder would be nice.
[15:45:30] <lerman> I'm thinking of using QT for the gui.
[15:45:52] <SWPadnos> do the subroutines have to live in the same "source" file, or can external files be called (ie, can there be a library of functions)
[15:46:05] <SWPadnos> QT is nice, but there are details that make it a PITA
[15:46:10] <lerman> It looks nice, I'm familiar with it, and is free (for non-commercial use).
[15:46:20] <lerman> What don't you like.
[15:46:56] <SWPadnos> I had started on a QT set of components, but I was (and still am) missing some bit of knowledge on how to make the components usable in QT Designer (for others)
[15:47:09] <SWPadnos> there is a QT interface to emc, but it's not complete
[15:47:20] <SWPadnos> (providing signals and slots)
[15:47:34] <SWPadnos> hey fenn, reboot complete? ;)
[15:47:52] <fenn> had a bagel while waiting to reboot
[15:48:03] <lerman> Where can I find the QT interface?
[15:48:36] <SWPadnos> I'll see if there's a copy somewhere on the net. else, I'll boot up my emc dev machine and upload it somewhere
[15:48:42] <lerman> fenn: either that was a small bagel, your machine is very slow to boot, or you really wolfed that down.
[15:48:54] <SWPadnos> Paul_c has a CVS repository, but I don't remember the IP (and I think you need a login)
[15:49:12] <Jacky^> SWPadnos: seen the video ?
[15:49:15] <lerman> Thanks. Can you email me a pointer to it (lerman-emc@se-ltd.com), please?
[15:49:17] <SWPadnos> (it's been an hour - it must be a big bagel)
[15:49:24] <SWPadnos> I haven't watched it all yet
[15:49:32] <Jacky^> was mk77 not napalm, but same effect
[15:49:35] <lerman> I've got to get to work. Bye.
[15:49:37] <SWPadnos> when I find it, I will
[15:49:39] <SWPadnos> see ya
[15:50:22] <fenn> well i sorta took a nap too
[15:50:32] <Jacky^> yeah..
[15:50:33] <SWPadnos> heh - sleeping of fthe bagel
[15:53:56] <SWPadnos> OK - time to really wake up. I'll be back soon
[15:53:59] <SWPadnos> SWPadnos is now known as SWP_Away
[16:43:10] <alex_joni> evening all
[16:43:22] <SWP_Away> SWP_Away is now known as SWPadnos
[16:43:31] <SWPadnos> hi Alex
[16:43:43] <alex_joni> hey stephen
[16:44:19] <Jymmm> ug
[16:44:26] <SWPadnos> oog
[16:44:29] <alex_joni> ug, Jymmm
[16:44:51] <fenn> is there any reason why halcmd can't create threads?
[16:45:18] <alex_joni> fenn: it's not implemented yet
[16:45:30] <alex_joni> but the main problem is that you create threads in the kernel
[16:45:48] <alex_joni> so you need to do that at insmod time.. can't communicate to that part of halcmd from userspace
[16:46:02] <alex_joni> if you look at the hal-rework, there is some work done there
[16:46:13] <alex_joni> which permits talking through /proc/hal/foo
[16:46:17] <fenn> eh, where's that?
[16:46:23] <alex_joni> so those limitations are stopped
[16:46:29] <alex_joni> it's a branch in emc2
[16:46:41] <fenn> cvs is down :(
[16:46:56] <fenn> * fenn mopes
[16:46:57] <alex_joni> http://cvs.sourceforge.net/viewcvs.py/emc/emc2/ works
[16:47:11] <alex_joni> halrefactor-0-1 is the branchname
[16:51:17] <alex_joni> fenn: tried a doze software today
[16:51:33] <alex_joni> does some very funky driver sh*t o windows, so it runs realtime
[16:51:39] <alex_joni> and it had a meter for latency
[16:51:48] <alex_joni> showed about 1-2 usecs latency
[16:52:01] <fenn> huh?
[16:52:09] <alex_joni> but when I started a program.. it went up to 1400 usecs :(
[16:52:09] <fenn> i thought that was processor dependent
[16:52:22] <alex_joni> so.. not useable
[16:52:44] <fenn> i'm thinking max_latency
[16:53:16] <alex_joni> 1.5 msecs :D
[16:53:39] <fenn> 1500000ns
[16:53:59] <fenn> is that any better than regular windows?
[16:54:14] <alex_joni> lol.. probably
[16:54:17] <alex_joni> not sure ;)
[16:56:01] <SWPadnos> "regular" windows can stall for seconds
[16:56:09] <SWPadnos> just get an IDE error
[16:56:16] <SWPadnos> (or something like it)
[17:09:48] <anonimasu> ;)
[17:09:57] <anonimasu> I've just tried the mill now
[17:14:58] <anonimasu> hand machining isnt too hard if you have a good machine :)
[17:16:08] <fenn> depends what you're doing i guess
[17:17:22] <anonimasu> fenn: a shitty machine wont make any parts..
[17:17:23] <anonimasu> :)
[17:17:39] <fenn> that's where you're wrong
[17:17:50] <fenn> i make all my stuff on a shitty machine :)
[17:18:01] <anonimasu> fenn: my old machine couldnt cut alu.
[17:18:15] <anonimasu> it should have been able to..
[17:18:17] <fenn> i find that hard to believe
[17:18:29] <anonimasu> but, it's not..
[17:18:34] <fenn> i use woodworking tools on aluminum sometimes
[17:18:50] <anonimasu> it would cut it but it chattered too badly to get any surface finish..
[17:19:16] <fenn> what was wrong with it?
[17:19:18] <anonimasu> and the spindle bearings would let the tool lift and then it chewed down into the alu and stuck..
[17:19:29] <anonimasu> it were a chineese mill..
[17:19:30] <anonimasu> :)
[17:19:37] <anonimasu> that about concludes it..
[17:19:38] <Jymmm> anonimasu: how deep a cut?
[17:19:43] <anonimasu> Jymmm: 2mm..
[17:19:56] <Jymmm> anonimasu with what typ/shape of bit?
[17:20:06] <anonimasu> 3 flute coated endmill..
[17:20:13] <anonimasu> 6mm dormer..
[17:20:30] <anonimasu> it'd work, but far from satisfactory..
[17:20:48] <Jymmm> I wonder if chipload was too high
[17:20:54] <Jymmm> being 3 flute
[17:21:02] <anonimasu> why would the chipload be too high?
[17:21:15] <Jymmm> anonimasu thats what I just asked
[17:21:21] <anonimasu> the feed and speed were right..
[17:21:35] <Jymmm> solid or tipped?
[17:21:43] <anonimasu> tipped..
[17:21:53] <anonimasu> dormer carbide are like 55$
[17:22:06] <Jymmm> Weird. Just like the $30 bits specifically for plastic I have to order.
[17:22:26] <anonimasu> yep..
[17:22:27] <Jymmm> http://onsrud.com/Products.asp?Action=View&Part=65-020
[17:23:05] <Jymmm> My only bitch about these bits is I'm not sure they'll resolve the chipload issue I'm having, though the mfg says it will.
[17:23:16] <anonimasu> hm what kind of issue?
[17:23:33] <Jymmm> HELP ME! I'm m e l t i n g
[17:23:38] <anonimasu> ah..
[17:23:50] <anonimasu> are you feeding too slow?
[17:23:58] <anonimasu> slowly..
[17:24:21] <Jymmm> See, these are 1) single flute and 2) have a special rack angle specicifally for plastic called an 'O' flute
[17:25:03] <Jymmm> So in "soft plastics", it'll curl the chips instead of allowing it to reweld itself
[17:25:05] <anonimasu> yes, but if you are feeding too slowly no voodoo can help you
[17:25:20] <Jymmm> anonimasu 8K rpm and 120 is the limit for me
[17:25:20] <anonimasu> *grins*
[17:25:32] <anonimasu> can you limit your rpm?
[17:26:02] <Jymmm> I just said that..... listen up! 8K rpm is the minimum I cna do
[17:26:16] <anonimasu> fenn: 8K rpm and 120 is the limit for me
[17:26:18] <anonimasu> whoops
[17:26:23] <anonimasu> I thought that was your max..
[17:26:35] <Jymmm> 8K RPM and 120IPM
[17:26:46] <Jymmm> 8k to 25K RPM is the range
[17:26:52] <anonimasu> what's the recomendation of the cutter mfg?
[17:26:55] <anonimasu> and the plastic mfg?
[17:26:57] <Jymmm> So, I'm going to different tooling
[17:27:26] <Jymmm> anonimasu this is cst and extruded acrylic
[17:27:29] <Jymmm> cast
[17:27:30] <anonimasu> tried compressed air to blow the chips away?
[17:27:53] <Jymmm> there are no "chips", it curls
[17:28:06] <anonimasu> ah..
[17:28:40] <Jymmm> this special plastic bits are to deal with that and pull the curls away from the material till it eventually breaks away
[17:28:45] <anonimasu> fenn: I kind of stopped beliving in that machine after I found out the taper were fscked /
[17:28:59] <anonimasu> have you tried them?
[17:29:41] <Jymmm> anonimasu: I talked to the mfg T/S (they have like 100+ plastic bits), and deciding which one I want to buy now.
[17:29:52] <anonimasu> ok
[17:30:14] <anonimasu> brb bath..
[17:30:27] <Jymmm> use soap and lots of it!
[17:30:27] <anonimasu> * anonimasu dosent like having to fear when he pushes run
[17:30:29] <anonimasu> yeah
[17:30:36] <anonimasu> *chews on the soap bar*
[17:30:53] <Jymmm> * Jymmm delouses anonimasu
[17:31:43] <Jymmm> AND... if I want a 1/8" bit witht hte shank taken down for deeper cuts, it's an extra $10 per bit
[17:32:13] <Jymmm> but, they'll at least do it.
[17:32:36] <fenn> it has less material in it so it costs more, duh.
[17:32:45] <Jymmm> lol
[17:32:48] <fenn> that's why laptops cost more than desktops
[17:33:05] <Jymmm> same with diamonds -vs- coal, right?
[17:33:10] <fenn> yeah
[17:33:12] <Jymmm> ok
[17:34:10] <Jymmm> * Jymmm hits the shower so he can remove the license plates and measure the bastards
[17:34:38] <fenn> stealing motorcycles again?
[17:35:27] <Jymmm> I wish. Just have an order for 4 signs to fit street legal golf carts
[17:35:43] <fenn> that sounds pretty lame
[17:35:54] <Jymmm> why's that?
[17:35:59] <fenn> i mean, money is cool, but golf carts aren't
[17:36:19] <Jymmm> Ha.... these are for a retirement community...
[17:36:36] <Jymmm> fuck walking so to speak, get a golf cart
[17:37:41] <Jymmm> $40 a sign @ 150 golf carts.... you do the math
[17:37:48] <fenn> wow
[17:38:03] <Jymmm> not so lame now, is it?
[17:38:16] <fenn> how big are the signs?
[17:38:24] <Jymmm> license plate sized =)
[17:38:45] <fenn> how did you get this job?
[17:38:56] <Jymmm> refereal
[17:39:42] <fenn> maybe you should get a decent spindle, or at least one that goes the right speed
[17:40:28] <Jymmm> I do have one. But i play with all kinds of materials, and acrylic is just tough.
[17:40:41] <Jymmm> it's just fine for the golf signs
[17:40:45] <Jymmm> ok, shower time
[18:17:16] <anonimasu> iab
[18:17:34] <alex_joni> ooohhh.. shiny
[18:17:36] <alex_joni> ;)
[18:17:42] <anonimasu> lol
[18:17:46] <anonimasu> alex_joni: what is?
[18:18:12] <alex_joni> you are :D
[18:18:35] <alex_joni> nice and clean, almost new =))
[18:18:40] <anonimasu> ah yeah
[18:18:41] <anonimasu> :D
[18:18:43] <anonimasu> *shines*
[18:18:52] <Jymmm> now to chrome plate anonimasu
[19:07:33] <les_w> I just ordered 150 pounds of that aluminum...if it looks good i'll order a thousand!
[19:11:04] <SWPadnos> have you ever heard of "Fortal" aluminum?
[19:11:44] <alex_joni> only "Fortran" aluminum
[19:11:48] <SWPadnos> heh
[19:12:10] <SWPadnos> it's a 7075-like alloy, but much higher yield strengths
[19:12:26] <SWPadnos> how about "Fortransit" - the precursor to Fortran
[19:12:26] <les_w> I have about 100 pounds of 7075
[19:13:01] <SWPadnos> take a look here: http://www.mousebar.com/fcchart.html
[19:13:13] <les_w> my aluminum supplies are dropping though. the remnants are getting smaller and smaller
[19:13:16] <les_w> looking
[19:13:40] <SWPadnos> there are other spots to look for info. I found it on eBay first
[19:14:53] <SWPadnos> ah - the technical manual: http://www.mousebar.com/fortmanl.html
[19:15:21] <les_w> good stuff
[19:15:34] <SWPadnos> how does it compare to MIC-6?
[19:15:53] <les_w> tooling plate is not strong, but very free from stress
[19:16:03] <les_w> let me check
[19:16:10] <SWPadnos> ah - ok
[19:16:31] <SWPadnos> I was looking for material to build motor mounts from, and I wanted something easier to machine than steel
[19:16:43] <SWPadnos> (since the machine is manual until the mounts are built :) )
[19:19:41] <les_w> Mic-6 is only about 20k psi yield. It's fine for light structural stuff where ease of machining and lack of warping is an issue.
[19:19:57] <les_w> motor mounts would be ok.
[19:20:10] <SWPadnos> for a Bridgeport-sized machine?
[19:20:39] <les_w> It's cast rather than wrought, so it makes fine chips. Cuts like butter.
[19:20:43] <SWPadnos> (actually, a Bridgeport)
[19:20:52] <SWPadnos> yep
[19:21:03] <les_w> Motor mounts? Yeah I think so.
[19:21:11] <SWPadnos> I bought some 7075-T651 on eBay, and that seems to cut pretty well
[19:21:26] <SWPadnos> It's also pretty hard
[19:21:29] <SWPadnos> (strong)
[19:21:36] <les_w> I use that (mic-6) for stuff not stressed a lot.
[19:21:50] <les_w> 6061 if I am gonna anodize it
[19:22:04] <les_w> 2024 or 7075 if it is highly stressed
[19:22:25] <SWPadnos> yeah - I suppose the stress on the mount isn't too high, but I was thinking about the mounting on a manual Bridgeport
[19:22:50] <SWPadnos> though I'm just realizing that it's the bolts that really matter
[19:23:02] <les_w> Ease of machining is the issue for me.
[19:23:05] <SWPadnos> the mount plate is just sandwiched between the bolt shoulders and the table
[19:23:10] <les_w> Like steels...
[19:23:32] <les_w> not stressed much...12L14 for sure.
[19:23:43] <les_w> welded...1018
[19:24:03] <les_w> higher stress...1144
[19:24:09] <les_w> then the tool steels
[19:24:24] <SWPadnos> yeah - what's with those? :)
[19:24:31] <SWPadnos> A vs D, etc
[19:24:36] <les_w> 12L14 is the choice if at all possible
[19:24:59] <les_w> tool steels? all different types
[19:25:09] <les_w> A is air hardening
[19:25:13] <SWPadnos> ah
[19:25:13] <les_w> O oil
[19:25:17] <les_w> etc
[19:25:31] <les_w> S shock resisting
[19:25:43] <SWPadnos> one day, I should read some of the introductory sections in Machinery's Handbook :)
[19:25:58] <les_w> D...forgot. Let me check....
[19:26:21] <SWPadnos> Deflection resistans? :)
[19:26:24] <SWPadnos> resistant
[19:26:52] <les_w> abrasion-toughness
[19:27:00] <les_w> D is for die
[19:27:03] <SWPadnos> ah
[19:27:43] <les_w> also known as hi carbon hi chrome
[19:27:47] <les_w> and it's nasty
[19:27:59] <SWPadnos> how?
[19:28:05] <SWPadnos> hard (to machine)?
[19:28:31] <les_w> yeah. not as bad as stainless or ... (jymmm) acrylic
[19:30:22] <les_w> anyway I was on autocad designing a new spindle bracket and dust collector shroud...Mic-6 is fine for that.
[19:30:40] <les_w> making a shroud that floats on thopson bearings
[19:30:53] <SWPadnos> yep, plus you're right - sands prices were pretty darned good - almost the bulk metal pricing
[19:31:09] <les_w> cna' beat it
[19:31:14] <les_w> can't
[19:31:24] <SWPadnos> (I thought you were being Scottish)
[19:31:34] <les_w> mic-6 is EXPENSIVE
[19:31:40] <les_w> it's good stuff
[19:31:48] <les_w> saves a lot of time too
[19:31:55] <SWPadnos> yeah - I saw some on eBay a while ago, and MSC is unaffordable
[19:32:31] <les_w> I used tons and tons of it for screen print vacuum plates
[19:32:33] <SWPadnos> or was that McMaster?
[19:32:46] <les_w> either...it's high
[19:33:27] <les_w> I am not sure how mic-6 would anodize...I'll assume poorly
[19:33:41] <SWPadnos> you can always try with a remnant
[19:33:47] <les_w> yep
[19:33:49] <SWPadnos> but I think it looks "flaky"
[19:34:01] <SWPadnos> sort of like galvanized roof is my bet
[19:34:01] <les_w> let me check
[19:35:10] <les_w> oh...it anodizes ok!
[19:35:12] <les_w> great
[19:35:19] <SWPadnos> cool
[19:35:36] <SWPadnos> I was thinking of setting up an anodizing bath, but hten I woke up ;)
[19:35:48] <les_w> well mic-6 has swirls from the blanchard gringing
[19:35:52] <SWPadnos> realized there were too many other things to do
[19:36:06] <SWPadnos> polish, then anodize
[19:36:07] <les_w> I bought all the stuff, but have not set it up yet
[19:36:12] <SWPadnos> make pretty colored mirrors
[19:36:50] <les_w> I am going to try it for a prototype woodworking tool I am making for someone...purple anodize!
[19:37:05] <les_w> practice on scraps first.
[19:37:07] <SWPadnos> heh - throw in some orange speckles, just for fun :)
[19:37:14] <les_w> yeah
[19:39:20] <les_w> Well, supposed to be off today...guess i'll attack some leaves
[19:39:29] <SWPadnos> heh - that's not vacation
[19:40:09] <les_w> aw it's fun..they are so dry a muching mower deck just turns them into powder
[19:40:17] <les_w> good fertilizer!
[19:40:20] <SWPadnos> that's our trick as well
[19:40:27] <SWPadnos> raking 1/2 acre by hand is a pain
[19:40:35] <SWPadnos> as is mowing :)
[19:40:59] <les_w> so is 5 acres witha couple hundred trees!
[19:41:14] <les_w> bur really...I like to mow. A lot.
[19:41:15] <SWPadnos> heh - I'm hoping you have a riding {mower, blower}
[19:42:17] <les_w> I have a gravely walk behind, Lawn tractor, and diesel front end loader.
[19:42:26] <les_w> So i'm covered
[19:42:35] <SWPadnos> OK - something more than a John Deere walk-behind
[19:42:39] <les_w> the FEL has mowers too
[19:42:51] <SWPadnos> you could be a golf course groundskeeper
[19:43:02] <SWPadnos> i'd just make one big sand trap
[19:43:15] <les_w> I will be...doing two holes in the back. 400 yards.
[19:43:52] <SWPadnos> well have fun. I suppose I should clean a desk so I can put the new computer on it
[19:44:19] <les_w> heh. No work for me. I should not have even been on autocad.
[19:44:21] <les_w> later
[19:44:27] <SWPadnos> see ya
[19:45:08] <les_w> les_w is now known as les_w_away
[19:52:54] <SWPadnos> SWPadnos is now known as SWP_Away
[19:56:07] <jepler> did the behavior of "override limit" change between emc1 and emc2? This message from Chatchai Neanudorn says that it's important to be able to "override limit" while the machine is not even turned on, but with emc1/simulator, the button immediately turns off again if the machine is in ESTOP or ESTOP RESET.
[20:10:40] <cradek> jepler: in tkemc, the button stays red
[20:10:46] <cradek> jepler: I have no idea whether or not that means something
[20:13:48] <cradek> jepler: [emc1 sim tkemc]
[20:15:21] <jepler> cradek: Does it ever turn not-red?
[20:15:30] <cradek> yes, when you finally jog
[20:15:58] <jepler> def toggle_override_limits(*args):
[20:15:59] <jepler> if not manual_ok(): return
[20:16:04] <jepler> oh, I bet I'm not even issuing the command
[20:16:14] <cradek> oops
[20:17:09] <lerman> Damn: "how about "Fortransit" - the precursor to Fortran" I was going to make a comment about Fortransit. It wasn't really a precurseor as much as an implementation. On the IBM 650, we used to punch our Fortran programs into cards. Then you would load the the Fortran compiler into the machine (perhaps a few thousand cards). So, you would get on line behind everyone else and put your deck...
[20:17:10] <lerman> ...into the card reader. Your program would be compiled and the resultant "IT" program would be punched out onto some more cards. (Or perhaps you would get some errors, instead.) Your IT program would be somewhat larger than the original Fortran deck. Then the IT translator would be loaded into the computer (a thousand or more additional cards). Get back on line and your IT program would...
[20:17:12] <lerman> ...be translated into SOAP (symbolic optimal assembly language). Or perhaps you would have some new errors. This deck would be very large. Next, the SOAP program would be translated into machine language. I don't believe you could get new errors at this point. Finally you would run your program -- it would crash and you went back to the beginning. Ahhh. The good old days.
[20:18:29] <SWP_Away> and hope that the lackey with the dolly-load of cards doesn't trip :)
[20:19:25] <jepler> I'm happier with modern PCs, thanks very much
[20:19:47] <SWP_Away> like the Altair - lots of toggle switches and lights on the front panel :)
[20:36:36] <K`zan> * K`zan was an IMSAI fan :)
[20:36:43] <K`zan> is still :)
[20:37:24] <lerman> At 10:10 (EST) today:
[20:37:26] <lerman> lerman:Where can I find the QT interface?
[20:37:27] <lerman> SWPadnos:I'll see if there's a copy somewhere on the net. else, I'll boot up my emc dev machine and upload it somewhere.
[20:37:28] <lerman> Any luck?
[20:39:17] <jepler> cradek: emcmodule doesn't let me pass any axis number but 0 in the EMC_AXIS_OVERRIDE_LIMITS message, so if we want SIMPLEINSTALL to work then we'll also have to live with my "set AUTO mode" hack
[20:39:41] <cradek> darn.
[20:40:29] <cradek> it's too bad about BDI.
[20:41:44] <SWP_Away> sorry - haven't had the chance to look for it. give me a little while for that
[20:42:09] <SWP_Away> BTW- it's a work in progress, I son't remember where it was left
[20:42:16] <SWP_Away> don't
[20:43:31] <lerman> OK. But a work in progress is a start. (As I used to say when I played bridge, "A peak at the cards is worth two finesses.")
[20:44:40] <SWP_Away> hw rubber of you :)
[20:44:42] <SWP_Away> how
[20:44:51] <SWP_Away> I'm booting the machine now - it takes a mo
[20:45:28] <SWP_Away> hmmm - this machine has USB - maybe I should find out where :)
[20:46:46] <lerman> Actually, the part that I'm interested in is the NML interface. I need to know how to 'talk' to the rest of EMC. How to give it a program to run, how to determine its present XYZ, etc. And I need a C (or C++) interface to it.
[20:47:18] <SWP_Away> this stuff exports signals and slots that QT widgets can use
[20:47:57] <SWP_Away> I also have some widgets (1-axis DRO) and things, but I'm not sure of the installation procedure
[20:49:34] <jepler> Can't jog axis 0 further past max soft limit.
[20:49:47] <jepler> cradek: is this the error?
[20:50:07] <cradek> I'm not sure
[20:50:22] <cradek> I don't think so
[20:50:23] <jepler> I figured out how to get it, and axis does freak out
[20:50:29] <cradek> oh, good
[20:51:03] <jepler> I set my X axis limits to not include zero, so home followed by any jog triggers it
[20:58:33] <SWP_Away> Hi Alex
[20:58:53] <jepler> ouch! axis segfaulted!
[20:59:05] <cradek> arg
[20:59:06] <SWP_Away> alex_joni, do you remember if anything else happened with the QT stuff since March/April?
[20:59:12] <Jacky^> mm
[21:00:26] <jepler> what's this mean? my emc prints it a lot once I get it in a certain state: 00000000 00000000 00000000 11000011
[21:00:39] <cradek> I think that prints when digital IO bits change
[21:00:43] <alex_joni> SWP_Away: what QT stuff?
[21:00:52] <cradek> toggle some of the lube and brake thingies
[21:00:56] <alex_joni> jepler: some weird debug
[21:00:58] <SWP_Away> qthmi
[21:01:03] <cradek> those are the emulated IO bits I think
[21:01:03] <jepler> but the numbers are the same each time!
[21:01:07] <jepler> no bit is changing
[21:01:12] <alex_joni> SWP_Away: what qthmi?
[21:01:16] <alex_joni> * alex_joni knows nothing :(
[21:01:18] <jepler> great, this time gdb segfaulted
[21:01:31] <SWP_Away> hmmm - I thought it was you (in on the big secret :) )
[21:01:32] <alex_joni> emc is known to break any software
[21:01:33] <alex_joni> :)
[21:01:35] <jepler> (gdb) continue
[21:01:35] <jepler> Continuing.
[21:01:35] <jepler> Program exited normally.
[21:01:36] <jepler> Segmentation fault
[21:01:48] <alex_joni> SWP_Away: I know anonimasu talked about some QT
[21:01:56] <alex_joni> and paul_c did some qt widgets
[21:02:01] <alex_joni> but that's all I know of...
[21:02:10] <SWP_Away> OK - maybe it was him (all you 'A' people - so confusing)
[21:03:25] <jepler> ugh -- it even prints with 'debug = 0' in ini
[21:07:51] <SWP_Away> actually - it was Paul's widgets I was referring to
[21:13:27] <SWP_Away> lerman - can you accept DCC?
[21:13:43] <lerman> What is DCC?
[21:13:57] <Jymmm> SWP_Away: That would be a no =)
[21:14:14] <SWP_Away> direct something or other - file transfer through IRC
[21:14:47] <SWP_Away> OK - you can, if it's not blocked by a firewall - hold on a sec
[21:15:19] <alex_joni> hmm.. did you guys get any ballots?
[21:15:25] <SWP_Away> nope
[21:15:42] <alex_joni> :/
[21:16:24] <lerman> gotta go. I have a business meeting right now.
[21:27:31] <dmess> hello all... ;)(
[21:27:40] <alex_joni> bonne soir
[21:27:45] <alex_joni> ca va?
[21:28:14] <dmess> pa t'elle mant
[21:29:04] <alex_joni> pour qouis?
[21:29:38] <dmess> la femme et les jeunes sont partit a la police's
[21:30:16] <SWP_Away> where's da po-lice?
[21:30:18] <alex_joni> a la police's? c'est tres graves?
[21:30:19] <Jymmm> Silly question... if you have an object in your hand and it has a radius, how would you measure that?
[21:30:30] <SWP_Away> with calipers?
[21:30:40] <SWP_Away> or a tape measure around, then divide by pi
[21:30:45] <dmess> mon plus vieux.. micrometer
[21:31:03] <Jymmm> measure what now?
[21:31:10] <alex_joni> dmess: mon francais et un peux.. rusty ;)
[21:31:13] <Jymmm> In this case, it's a license plate.
[21:31:15] <dmess> how big... pi tape works with a skilled operatol
[21:31:16] <SWP_Away> or, if it's flat, use a tape measure across the diameter
[21:31:28] <SWP_Away> ah - a radius on a non-circular item
[21:31:40] <Jymmm> Yeah, just the corners
[21:31:42] <alex_joni> Jymmm: you mean rounded corners?
[21:31:45] <dmess> cmm
[21:31:53] <alex_joni> Jymmm: mark the points where the lines turn to circles
[21:32:01] <alex_joni> and measure the distance between those
[21:32:05] <SWP_Away> yep
[21:32:07] <alex_joni> should be the diameter
[21:32:13] <dmess> or a dowwel hole in the same setup to mic to
[21:32:17] <alex_joni> radius is half ;)
[21:32:27] <SWP_Away> and if this is for license plates, look at it, then estimate to the nearest 1/16 inch - it'll be close enough
[21:32:51] <SWP_Away> (I'm betting 3/8")
[21:32:54] <SWP_Away> maybe 1/4
[21:32:55] <dmess> program cut and SHIP
[21:33:11] <SWP_Away> make them square, then use a belt sander :)
[21:33:25] <dmess> blanking licence plates???
[21:33:33] <SWP_Away> non-license plates
[21:33:50] <SWP_Away> for old farts in those Chrysler street-legal golf carts
[21:34:04] <Jymmm> SWP_Away you're just jealous
[21:34:14] <dmess> send to wire cutter for 250.00 and itll be pristene
[21:34:44] <dmess> punch and die outa the same block..
[21:35:28] <dmess> cnc form them and it a i hit wonder on a 1 ton mech press
[21:37:43] <dmess> material??
[21:38:11] <SWP_Away> acrylic, I think
[21:38:18] <SWP_Away> (based on earlier discussions)
[21:42:56] <dmess> benching takes time.... if you can cut the acrylic thats betterbenching IS not good for YOU
[21:44:38] <dmess> any graphics??
[22:21:50] <lerman> SWP_Away: lerman is back. Do you want to try that DCC, again?
[22:21:56] <SWP_Away> sure
[22:22:32] <SWP_Away> how's that?
[22:22:44] <lerman> I seem to have it.
[22:22:48] <SWP_Away> cool.
[22:23:05] <SWP_Away> I just tarred the whole tree, so you have backup files and CVS info in there as well
[22:23:24] <SWP_Away> you can't cvs up, so you could just remove the CVS dirs
[22:23:31] <lerman> It untarred without error.
[22:23:34] <SWP_Away> cool
[22:23:57] <lerman> * lerman says thank you.
[22:23:59] <SWP_Away> there are a couple of .ui projects, as well as some C (cc, and hh actually) source
[22:24:05] <SWP_Away> you're welcome
[22:24:20] <SWP_Away> and thank paul_c for the work - it's mostly his (almost all of it, actually)
[22:26:25] <alex_joni> * alex_joni goes to bed..
[22:26:27] <alex_joni> night guys
[22:26:32] <SWP_Away> noght
[22:26:33] <Jymmm> G'Night alex_joni
[22:26:43] <SWP_Away> i - n*i*ght
[22:31:10] <SWP_Away> lerman, the thing I had a hard time with was making QT designer (or kdevelop) UI elements that someone else could then drop into their own UI later (like the installable components in C++ Builder and the like)
[22:31:35] <SWP_Away> I just never quite "got" how to do it
[22:31:58] <SWP_Away> SWP_Away is now known as SWPadnos
[22:32:28] <SWPadnos> making a usable component was my goal, not so much the actual UI demo that's there (which should work, by the way)
[22:41:02] <lerman> My intention is to NOT make a component that is reusable in the QT designer sense. I'd like to make a single component that is reusable. It would be configurable from a file. The configuration file would specify a background image (or set of images), a set of labeled integer entry blocks with their locations, possibly some selectors for modes (climb vs conventional), etc. The purpose of...
[22:41:04] <lerman> ...this is to let the user create his own wizards (in the sense that mach has them). So, for a rectangular pocket wizard, the background would have a drawing of a pocket with pointers to things representing depth, width, height, pass width, cutter diameter, etc. Integer input widgets would be positioned so that it was clear what they represented. I think this should be straightforward to do.
[22:41:46] <SWPadnos> likely more straightforward than a QT designer component
[22:42:05] <SWPadnos> I wanted to get away from the full-blown GUI though
[22:42:26] <SWPadnos> some people run lathes (2 axes), others hexapods (with 6-axis jogging needed)
[22:42:36] <lerman> Well, in a past life I did make a few QT designer widgets. And then promptly stopped using them because they were more effort than they were worth.
[22:43:07] <SWPadnos> components always are, until you use them the second or third time
[22:43:16] <SWPadnos> assuming that you can
[22:44:01] <lerman> Well, someone who cares about lathes could/should build that gui. But most users are running mills of one type or another. In my experience, building good reusable components takes 10 times the effort of just building things.
[22:44:13] <SWPadnos> at least
[22:44:21] <SWPadnos> did you ever get any to work?
[22:44:27] <SWPadnos> (in QT)
[22:46:02] <lerman> Yes, I had some widgets working in QT. But it was easier to just build a monolithic screen than it was to change the widgets to do what I wanted. A shitload of little files may seem like a good idea until you have to manage them.
[22:46:38] <SWPadnos> sure - but note how easy it is to design a UI in QT designer
[22:46:53] <SWPadnos> (of course, using it in an actual program is a little harder)
[22:48:13] <lerman> Yup. So, draw the screen with all the existing widgets, draw your own widgets in the main control program without making them widgets.
[22:49:06] <SWPadnos> the point of the QT exercise was to have widgets that could be connected to EMC, and would "know" what to do when they were
[22:49:45] <SWPadnos> rather than having everyone who wants to make a 1-axis DRO have to write the same code to format position data into the right number of decimals, etc.
[22:50:25] <SWPadnos> and then add to that the ability to drop 3 (or 6) of these 1-axis DROs on the same screen, and end up with a 3-6 axis DRO when you're done
[22:50:32] <lerman> I understand. And I think it's a good idea. You don't need that many types of widgets.
[22:50:41] <SWPadnos> nope
[22:51:04] <SWPadnos> I was thinking about a single axis with DRO, mode, and jog (possibly optional on the jog)
[22:51:15] <SWPadnos> possibly a special button that takes state from a signal
[22:51:42] <SWPadnos> there's not much else that you need for EMC specifically (maybe an MDI widget)
[22:52:20] <lerman> Before, you mentioned jogging hexapods. But, do you really want to jog each actuator separately? Do we jog cartesian directions or joints?
[22:52:32] <SWPadnos> either, and yes
[22:52:51] <SWPadnos> you need to be able to get the individual joints homed, without them going through kinematics
[22:53:15] <SWPadnos> or zeroed initially (or after a crash / limit problem)
[22:55:13] <SWPadnos> also, a 4-axis cartesian machine isn't necessarily XYZA - a Bridgeport with quil land knee motorized may need to be individually jogged
[22:56:38] <lerman> Yes. Although my Bridgeport (clone) has a big handle that I have to crank to raise and lower the knee I'd like to put a motor on it some day to save the right arm.
[22:56:49] <SWPadnos> yep - mine too
[22:57:01] <SWPadnos> (though it does save me a membership in a health club
[22:57:02] <SWPadnos> )
[22:57:23] <lerman> I get a 50% discount -- my left arm still needs to go to the club.
[22:57:31] <SWPadnos> heh
[22:57:43] <SWPadnos> use that one to lift the large steel blocks onto the table
[22:57:55] <SWPadnos> (or the vise and/or rotary table)
[22:58:30] <lerman> I machine mostly aluminum -- but also have the vise to move. Why do you need a rotary table with CNC? (unless you turn it on its side)
[22:59:08] <SWPadnos> I don't have a CNC yet :)
[22:59:13] <SWPadnos> just motors and a machine
[22:59:24] <SWPadnos> (and no mounts for said motors)
[23:00:24] <lerman> I wound up getting an Elrod kit for all of that stuff. (I had more money than time and energy.)
[23:00:36] <SWPadnos> what was the cost, if I may ask?
[23:02:55] <lerman> I don't recall the details. But the pricing is on their web page. Perhaps $5000 for everything. Ball screws $1650, I think.
[23:03:04] <SWPadnos> hmmm - OK
[23:03:36] <SWPadnos> come to thikn of it, I think I spoke to Dwayne some time ago. It was something like $1600 for the mounts alone (and that may have been for scratch-n-dents)
[23:05:23] <lerman> The Z axis is quite nice. I bought the whole package. (But the mill I installed it on only cost me $250.)
[23:05:56] <SWPadnos> I actually got an Anilam Z drive kit on eBay (around $200 I think, no motor)
[23:10:11] <lerman> Dinner time. Gotta go. See you later. Ken
[23:10:16] <SWPadnos> see ya