#emc | Logs for 2004-12-07

[00:01:39] <jepler> cradek: hah
[00:29:49] <cradek> jepler: what are you trying to do?
[00:33:57] <cradek> hi jmk
[00:34:20] <jmkasunich> hi
[00:35:30] <jepler> cradek: Started on using the real interpreter instead of ours
[00:35:41] <cradek> jepler: right
[00:35:56] <cradek> jepler: how's that going?
[00:36:17] <cradek> what data type can you use to return the canon to python?
[00:37:00] <cradek> or are you going to do one line at a time like the c api?
[00:37:17] <jepler> I'm waffling
[00:37:49] <cradek> that's understandable
[00:37:57] <cradek> it would be nice to have one chunk, but it could be really big
[00:38:22] <jepler> I had something written that turned the important EMC_TRAJ_xxx messages into Python objects, but then I decided that I wanted to not use the code in emccanon.cc but to define STRAIGHT_FEED and a billion other functions and make them callbacks to a Python object
[00:50:18] <jepler> the main problem with emccanon.cc is that its output only has the idea of a single kind of linear move, with some calculated feed rate, which leads to the kind of guessing seen in emcplot3d
[00:52:12] <jepler> the main problem with talking directly to rs274ngc is that it doesn't take care of things like offsets before you see the coordinates, and it requires you to do a lot of bookkeeping
[00:54:02] <jepler> which brings me back to the reasons I wanted to write my own interpreter, except that I have no interest in doing all the canned cycles, or cutter radius compensation, or even arcs for that matter
[00:56:44] <jmkasunich> the danger in writing your own interpreter is making sure that it works _exactly_ like the existing one... plotting with one interp and cutting with another one seems risky to me
[00:57:56] <jepler> agreed
[00:58:21] <robin_sz> hi john ...
[00:58:49] <jmkasunich> hi robin
[00:58:54] <CIA-9> 03Zathras 07BDI build system * 10Babylon Cluster/comps.xml: File changed. New revision:1.16
[00:59:30] <robin_sz> hey, after several months of messing abut, my mate seems to have a decent switch mode PSU working, remember the one to drive some discharge lamps for a laser?
[01:00:07] <jmkasunich> yeah
[01:00:32] <robin_sz> jmkasunich: hes getting about 10kw out, only needs 6kw though
[01:00:47] <robin_sz> so it looks tasty :)
[01:00:49] <jmkasunich> that's not a problem is it?
[01:01:01] <robin_sz> nope, thats a good thing
[01:01:21] <jmkasunich> how efficient it it?
[01:01:34] <robin_sz> exact percentages? ...
[01:01:37] <robin_sz> dunno
[01:01:45] <jmkasunich> hopefully over 90%
[01:01:52] <robin_sz> runs fairly cool at 6kw,
[01:01:57] <robin_sz> yes, over 90 for sure
[01:01:59] <jmkasunich> well over, unless he has pretty good cooling
[01:02:17] <jmkasunich> the stuff I work on is 97-98%
[01:02:22] <jmkasunich> but higher power levels
[01:02:23] <robin_sz> * robin_sz nods
[01:02:28] <robin_sz> yeah, I know :)
[01:02:59] <robin_sz> using the semikron stuff, they have some neat devices
[01:03:30] <jmkasunich> the only semikron stuff I've ever used was SCRs, over 13 years ago
[01:03:48] <jmkasunich> they sell some IGBT modules with built in gate drives, don't they?
[01:03:53] <robin_sz> yeah
[01:04:02] <jmkasunich> that what your mate's using?
[01:04:22] <robin_sz> IGBT, own gate driver, some PWM chip
[01:04:31] <jmkasunich> not bad
[01:04:32] <robin_sz> 3526
[01:04:48] <jmkasunich> buck topology?
[01:04:54] <robin_sz> nope
[01:05:00] <jmkasunich> what then?
[01:05:31] <jmkasunich> * jmkasunich doesn't remember the input volts and desired output volts/amps
[01:05:47] <jmkasunich> constant current, wasn't it?
[01:06:12] <robin_sz> H bridge driving a txfmr
[01:06:12] <robin_sz> yep
[01:06:12] <robin_sz> 50A at 200V
[01:06:14] <robin_sz> ish
[01:06:37] <jmkasunich> rectifier on the txfmr output?
[01:06:39] <robin_sz> there seems to be a marketing niche for fully isolated designs
[01:06:41] <robin_sz> * robin_sz nods
[01:06:53] <jmkasunich> this is a product, or a one-off?
[01:07:10] <robin_sz> the major competitor uses a buck convertor, so being isolated is a plus point
[01:07:12] <robin_sz> well ...
[01:07:21] <robin_sz> it started off as a one off sor t of thing
[01:07:22] <jmkasunich> transformer design must have been interesting - I'm rather impressed
[01:07:36] <robin_sz> but then he saw a potential for it on the open market
[01:07:55] <robin_sz> I think his desing for the txmr was a bit 'empirical' :)
[01:08:18] <jmkasunich> nevertheless, at those power levels, a switching transformer design is non-trivial
[01:08:20] <CIA-9> 03Zathras 07BDI build system * 10Babylon Cluster/RELEASE-NOTES: File changed. New revision:ks.cfg
[01:08:27] <robin_sz> basically, he had a choice of two possible cores, got the bigger one and a lot of copper tape ...
[01:08:49] <jmkasunich> how big is the xfmr (cm, kg)?
[01:09:05] <robin_sz> say, 1kg
[01:09:33] <robin_sz> 140mm sides, 30mm square section
[01:09:47] <robin_sz> from memory
[01:09:48] <jmkasunich> is there an inductor in the circuit too (forward converter?)
[01:10:01] <robin_sz> yes,
[01:10:07] <jmkasunich> he make that too?
[01:10:07] <robin_sz> gapped silicon iron core
[01:10:11] <robin_sz> * robin_sz nods
[01:10:30] <jmkasunich> how fast does it switch, 10KHz, more, less?
[01:10:41] <robin_sz> 20khz
[01:10:48] <robin_sz> I think ...
[01:10:50] <jmkasunich> nice
[01:11:11] <jmkasunich> the xfmr design is certainly simpler for a forward converter, the energy storage is elsewhere
[01:11:37] <robin_sz> in the inductor, right
[01:11:48] <jmkasunich> and the H bridge clamps leakage inductance spikes, so it's not quite so difficult, but still I'm impressed
[01:11:56] <paul_c> cradek, jepler: python-axis added to the CD.
[01:12:39] <robin_sz> jmkasunich: hes got it all onto one board, 300mm long, 150mm wide, plus the magentics, obviously
[01:13:14] <jmkasunich> four igbts, four gate drivers, at least two isolated, xfmr and inductor, and at least two output diodes... I imagine it's a bit spendier than the buck competitor
[01:13:25] <robin_sz> nah
[01:13:30] <robin_sz> cheaper
[01:13:31] <jmkasunich> one igbt, one diode, one inductor
[01:14:16] <robin_sz> yes ... but the competitor somehow manages to use about 4 pcbs with various crap on them, a crazy hand wired loom, and a chassis that costs a fortune
[01:14:27] <robin_sz> this is one pcb, moslty surface mount
[01:14:56] <jmkasunich> so bad execution drives up the cost of the fundamentally simpler and cheaper solution
[01:15:01] <jmkasunich> nothing new there ;-)
[01:15:07] <robin_sz> heh :)
[01:15:15] <robin_sz> the semikron modules are neat
[01:15:19] <robin_sz> basically,
[01:15:43] <robin_sz> you bolt them to the heatsink with the PCB on top
[01:15:57] <jmkasunich> six packs?
[01:16:11] <robin_sz> pads on top of the device contact the PCB, no soldering
[01:16:17] <robin_sz> just bolt em down
[01:16:52] <jmkasunich> bolts into the top of the module (at each terminal), or just pressure on the board from external bolts?
[01:17:41] <robin_sz> * robin_sz cant remember
[01:17:41] <robin_sz> I thnik they are pairs
[01:17:41] <robin_sz> just presure I think
[01:17:59] <jmkasunich> * jmkasunich browses semikron.com
[01:18:02] <robin_sz> the top of the IGBT has jut one bolt hole, and 4 pads IIRC
[01:18:09] <robin_sz> I could be dreaming of course :)
[01:18:20] <robin_sz> ah no, one at each terminal
[01:19:23] <jmkasunich> duals you say... so +, -, and output power terminals, and two gates, two kelvin emitters for signal terminals
[01:19:25] <robin_sz> the 'semitop' stuff is it
[01:20:00] <robin_sz> right, no I was confusled
[01:20:20] <robin_sz> the pad thing must have been someting else
[01:20:42] <robin_sz> * robin_sz is easily confused
[01:20:58] <jmkasunich> so two modules for the H bridge
[01:21:05] <robin_sz> yeah
[01:21:17] <jmkasunich> more modules of some sort for the input rectifier and output rectifier?
[01:22:22] <robin_sz> yeah
[01:22:30] <robin_sz> 4 modules total I think
[01:22:35] <jmkasunich> sounds right
[01:22:56] <robin_sz> was trying to decide on
[01:23:04] <robin_sz> input bridge in diodes
[01:23:05] <robin_sz> or
[01:23:13] <robin_sz> half-contrlled SCR bridge
[01:23:14] <jmkasunich> one heatsink, 4 modules, one board... caps soldered to the board, only the magnetics is off-board
[01:23:21] <robin_sz> * robin_sz nods
[01:23:34] <jmkasunich> SCRs are elegant, but overkill IMO at 10Kw
[01:23:43] <jmkasunich> resistor shunted by a relay
[01:23:48] <robin_sz> wyeah
[01:23:56] <robin_sz> or a thryistor
[01:24:23] <jmkasunich> we do resistor/relay up to 100Kw+
[01:24:35] <robin_sz> yeah, but that needs wiries
[01:24:38] <robin_sz> wires
[01:24:57] <jmkasunich> what is input voltage?
[01:25:13] <robin_sz> something on-board, like a thryristor can be stuffed as the board is stuffed
[01:25:19] <robin_sz> 415V 3p
[01:25:48] <jmkasunich> we use PCB mounted relays and resistors on 50HP (35Kw) drives
[01:25:56] <robin_sz> right
[01:26:23] <jmkasunich> put the relay in series with the cap, so it only carries the ripple current, not the full load current
[01:26:51] <robin_sz> * robin_sz ponders that
[01:27:02] <robin_sz> hmmm
[01:27:06] <robin_sz> good point :)
[01:27:19] <jmkasunich> the difficulty is that the relay can add inductance between caps and IGBTs - that's not good
[01:27:27] <jmkasunich> but we get away with it at 50HP
[01:27:33] <robin_sz> * robin_sz nods
[01:27:47] <jmkasunich> (there is a small cap, 0.5-1.0uF located right on top of the IGBTs
[01:28:00] <robin_sz> yeah, helps
[01:28:20] <robin_sz> and a small R in the gate drive, right up close
[01:28:26] <jmkasunich> larger stuff uses either phase control, or three phase contactor shunting resistors on the input, before the rectifier
[01:28:41] <jmkasunich> switching AC is much easier on the contactor
[01:28:48] <robin_sz> he has a contactor at the moment, but thats to oexpensive for production
[01:29:26] <jmkasunich> dammit you've done it again... frittered away a half hour talking about electronics ;-)
[01:29:31] <robin_sz> basically, the more that can be placed automatically by the pick and plcae, the better
[01:29:35] <robin_sz> tee hee :)
[01:30:01] <jmkasunich> * jmkasunich needs to investigate an intermittent bug that stops slot 2 of the compile farm
[01:30:42] <robin_sz> weve got a Zevatech 730 pick and place that wll stuff even quite large components onto a board
[01:30:48] <jmkasunich> "investigage" means add some "echo"s to the script and wait a couple days for it to fail again
[01:31:18] <jmkasunich> "we" as in you and your mate?
[01:31:18] <robin_sz> heh
[01:31:24] <robin_sz> well, yes, sort of
[01:31:31] <robin_sz> its complicated :)
[01:31:38] <jmkasunich> isn't everything?
[01:31:50] <robin_sz> half his gear seems to live in my factory
[01:32:04] <robin_sz> and I seem to have keys to his factory
[01:32:18] <robin_sz> Im getting a bit confused as to who owns what anymore :)
[01:32:32] <jmkasunich> it only matters when somebody breaks something
[01:32:42] <robin_sz> yeah,
[01:32:52] <robin_sz> so far, thats not happened
[01:33:02] <jmkasunich> you must be hitting the big times now... it's a "factory", not a "workshop" now
[01:33:12] <robin_sz> heh, nearly .. oh so very nearly
[01:33:23] <robin_sz> distributors coming on wednesday ...
[01:33:39] <robin_sz> final appproval on the machine
[01:33:58] <jmkasunich> my mental image of your shop is still just a large garage or something, outside your home
[01:34:10] <robin_sz> 1500 sq feet
[01:34:34] <robin_sz> small indutrial unit about 5 miles from home
[01:34:42] <jmkasunich> I see
[01:35:02] <jmkasunich> about 3 or 4x the typical 2 car garage
[01:35:05] <jmkasunich> nice
[01:35:17] <jmkasunich> * jmkasunich would kill for 500 square feet
[01:35:23] <robin_sz> in the uk
[01:35:28] <robin_sz> a 2 car garage is ...
[01:35:36] <robin_sz> big enough for 2 cars
[01:35:41] <jmkasunich> 20 x 20?
[01:35:50] <robin_sz> and just enough room to get down the sides
[01:36:13] <jmkasunich> 15 x 15?
[01:36:21] <jmkasunich> 2 Mini's ;-)
[01:36:21] <robin_sz> thats more like it
[01:36:40] <jmkasunich> ours is 22w x 28 deep - I added the extra 8 feet
[01:36:49] <jmkasunich> normal is 22x20 or so
[01:36:53] <robin_sz> a typical UK garage gives you enough room for say a freezer and a car
[01:37:08] <robin_sz> but it is unlikely you can open the freezer with the car in as well
[01:37:16] <jmkasunich> the extra fills up with lawnmower, garden stuff, spare furniture, and whatnot
[01:37:26] <jmkasunich> even my extra 8 feet is full of crap
[01:37:36] <robin_sz> yip
[01:37:39] <jmkasunich> I have about 8x8 in one corner where the Van Norman mill is
[01:37:44] <robin_sz> rubbish expands to fill the space available
[01:37:57] <jmkasunich> I wish I had made it 40ft deep
[01:38:04] <robin_sz> ah yes
[01:38:13] <robin_sz> unlimited amounts of land, handy ;)
[01:38:14] <jmkasunich> the yard is big enough - I could have walled off the back half and heated it
[01:38:54] <jmkasunich> (It really sucks working with 1800lbs of iron that has been soaking in -5C for weeks
[01:39:05] <robin_sz> heh
[01:39:17] <robin_sz> I know ...
[01:39:22] <robin_sz> drag it into the house!
[01:39:27] <jmkasunich> prices for my side job customers rise in the winter
[01:39:32] <jmkasunich> no room in the hose
[01:39:34] <jmkasunich> house
[01:39:46] <robin_sz> sure there is ...
[01:39:49] <jmkasunich> I have about 16x8 in the basement, already full
[01:40:00] <jmkasunich> the steps couldn't handle the mill
[01:40:16] <robin_sz> just put it in the middle of the living room
[01:40:28] <robin_sz> it will find its way to tthe basement soon enough
[01:40:28] <jmkasunich> the floor couldn't handle it either
[01:40:51] <jmkasunich> more to the point, my wife couldn't handle it - I'd be out on my ear ;-)
[01:41:20] <robin_sz> heh
[01:41:38] <robin_sz> had to send the steel delivery guy away on his ear today
[01:42:08] <robin_sz> lorry loaded by fsckwits
[01:42:24] <robin_sz> I was first drop, 6 sheets of 3mm
[01:42:38] <robin_sz> it was under some sheets of 6mm ...
[01:43:00] <robin_sz> and in the middle of the wagon, across longways
[01:43:20] <robin_sz> so I coulnt even lift the 6mm off with the forklift
[01:43:47] <robin_sz> if it had been sideways on I would have been able to ...
[01:43:50] <robin_sz> idiots
[01:44:24] <robin_sz> * robin_sz hates steel stockists
[01:45:00] <paul_c> BDI-4.01 is ready for Ray to try and break.
[01:45:17] <jmkasunich> 4? decided to skip 3?
[01:45:21] <robin_sz> coo, big numbers
[01:45:33] <paul_c> TNG should have been -3.xx
[01:45:52] <robin_sz> are we doing even numbers for stable releases, odd for testing?
[01:46:01] <robin_sz> like kernels?
[01:46:03] <jmkasunich> ok, so 4.x is what you were calling 3.x a week ago?
[01:46:20] <paul_c> yup
[01:46:42] <paul_c> gone from alpha to beta-testing
[01:47:02] <robin_sz> coo
[01:47:09] <jmkasunich> hmmm, one of these days I gotta try that
[01:47:23] <paul_c> wot, beta-testing ?
[01:47:30] <jmkasunich> BDI4
[01:47:55] <paul_c> spare slot on the farm ?
[01:47:59] <jmkasunich> I know a guy at work with a CD burner, and work has something like T3 inet access
[01:48:04] <jmkasunich> that's what I was thinking
[01:48:15] <jmkasunich> will it run on 128MB, 200MHz?
[01:48:21] <paul_c> still take you 12 hours to download
[01:48:37] <jmkasunich> slow on your end?
[01:48:52] <paul_c> dialup speeds going out ;(
[01:49:11] <jmkasunich> goes out from where? your place?
[01:49:26] <paul_c> yup
[01:49:51] <jmkasunich> you have better than dialup (I thought)... or do you throttle it?
[01:50:19] <paul_c> 750K down, 128K up
[01:50:35] <jmkasunich> I see
[01:50:43] <robin_sz> hence asynchronus DSL
[01:51:00] <robin_sz> 750K down is good though
[01:51:03] <robin_sz> cable?
[01:51:11] <jmkasunich> is mike still hosting stuff for you>
[01:51:13] <jmkasunich> ?
[01:51:32] <paul_c> for the ISOs ?
[01:51:35] <jmkasunich> yeah
[01:51:45] <jmkasunich> actually, is mike (tbl) still there?
[01:51:56] <paul_c> dunno - Need to talk to him about that
[01:52:22] <robin_sz> he was at sherline, right>
[01:52:23] <robin_sz> ?
[01:52:27] <jmkasunich> yeah
[01:52:37] <robin_sz> they still do the emc powered mill?
[01:53:04] <paul_c> shifting 30+ a month
[01:54:33] <cradek> last I heard, mike left sherline
[01:57:24] <jmkasunich> so do you think 4.x will run on the farm, or do I need more grunt?
[01:58:34] <paul_c> KDE-3.2 is installed by default, but it doesn't run untill you log in
[01:58:59] <paul_c> 128Meg should be plenty.
[01:59:14] <robin_sz> 30 a month, neat business :)
[01:59:48] <jmkasunich> how would I avoid it starting?
[01:59:58] <paul_c> Don't log in
[02:00:08] <jmkasunich> smartass
[02:00:20] <paul_c> <smirk>
[02:00:21] <jmkasunich> how would I avoid it starting and be able to use the box from the command line?
[02:00:40] <jmkasunich> want to edit the default runlevel, or something like that
[02:00:50] <paul_c> delete kdm from rc2.d
[02:01:24] <paul_c> or swith to a tty
[02:01:51] <jmkasunich> ctrl-alt-F2, then login?
[02:02:01] <paul_c> F1 to F6
[02:02:15] <paul_c> kdm runs on tty7 only.
[02:02:31] <jmkasunich> well it will come up on F1, if I understand you logging in on F1 will automatically start KDE
[02:02:49] <jmkasunich> _then_ I guess I could switch to F2, and log in again?
[02:02:52] <paul_c> No - tty1 is a text only login
[02:03:05] <jmkasunich> * jmkasunich is thick tonight
[02:03:17] <jmkasunich> KDE-3.2 is installed by default, but it doesn't run untill you log in
[02:03:19] <paul_c> tty7 (or Alt-F7) for the GUI login
[02:04:07] <jmkasunich> so if I just boot it up, I'll see a text mode login prompt? or will it start X and present a GUI login prompt?
[02:04:14] <paul_c> KDE-3.2 is installed by default, but it doesn't run untill you log in on the GUI login screen found on F7
[02:04:33] <jmkasunich> I think I get it - it will start X and present the GUI prompt, but I can switch to F1 and login in text mode
[02:04:54] <jmkasunich> X will still be running tho, right? (but not KDE)
[02:05:07] <paul_c> It will default to the GUI login screen, but a simple Ctrl-Alt-Fn will get you a text screen.
[02:05:25] <jmkasunich> I understand now (told you I was thick)
[02:05:50] <jmkasunich> If I want to completely avoid X, then I have to go in and edit startup scripts
[02:06:21] <paul_c> Just remove the kdm symlnk frm /etc/rc2.d
[02:06:36] <paul_c> That will be enough.
[02:06:48] <jmkasunich> ok
[02:08:06] <paul_c> Off the top of my head, I think it is /etc/rc2.d/S99kdm
[02:08:35] <paul_c> leave the Kxxkdm symlink in place.
[02:08:56] <jmkasunich> by the time I get around to downloading it and burning the CD and doing the install, you'll be up to 4.3 or something, and I'll have forgotten most of this discussion
[02:09:20] <jmkasunich> the F1 method will probably stick with me, and I'll do that to get started
[02:09:39] <paul_c> One small problem though...
[02:09:53] <jmkasunich> there's always one of those
[02:10:07] <paul_c> EMC from SF will never compile on it
[02:10:21] <jmkasunich> _any_ emc? 1 or 2?
[02:10:32] <paul_c> emc2 tree is the only one that has any chance.
[02:10:33] <jmkasunich> oh, thats right - no dev tools
[02:10:49] <paul_c> Oh, dev tools are all in place.
[02:10:57] <paul_c> (at the moment)
[02:11:00] <jmkasunich> I thought they were gonna be on the 2nd CD
[02:11:29] <paul_c> By the time it gets to public release, maybe.
[02:13:37] <jmkasunich> in any case, it would be good to play around with it.
[02:14:10] <jmkasunich> let me know when you have it posted somewhere with some bandwidth (I don't want to use 12hrs of your outgoing)
[02:14:27] <jmkasunich> even if that doesn't happen till 4.3
[02:15:05] <jmkasunich> (how is Ray getting it anyway? he has even less bandwidth)
[02:15:48] <paul_c> Either though the local school, or someone in the town.
[02:16:23] <jmkasunich> maybe once he has it and has a few days testing on it, I'll send him some $ and he can burn me a disk
[02:16:48] <jmkasunich> (I'd send you $ and let you burn me one, but shipping is probably slow and maybe expensive)
[02:17:04] <jmkasunich> besides, you probably want those funny L's instead of $
[02:17:25] <paul_c> 'bout 68p to post one over - takes about a week.
[02:17:46] <jmkasunich> thats between $1 and $2?
[02:18:01] <paul_c> $1.23
[02:18:11] <paul_c> (and no sales tax)
[02:18:32] <jmkasunich> hmmm.... assuming there are no oopses in the first couple of days, can you send me one?
[02:18:52] <jmkasunich> I can either send you money, or pay for it in beer at NAMES
[02:21:34] <paul_c> There are still some issues to resolve
[02:21:35] <jmkasunich> scratch NAMES, make that Fest
[02:21:35] <paul_c> The only way you'll get me at NAMES is if someone pays me to go.
[02:21:35] <jmkasunich> that was a mental type
[02:21:35] <jmkasunich> typo
[02:21:35] <jmkasunich> what about Fest?
[02:21:35] <paul_c> 'pends when, where, and if.
[02:21:35] <jmkasunich> or are you swearing off the states completely for a while
[02:21:35] <asdfqwega> * asdfqwega wounldn't go to NAMES again, either
[02:21:35] <jmkasunich> same here
[02:21:35] <jmkasunich> I _think_ the where will probably be Roland's place
[02:21:35] <paul_c> 'bout the only reason for going to NAMES would be to bait Fred Smith.
[02:21:35] <asdfqwega> ?
[02:21:35] <jmkasunich> when either week after or week before names
[02:21:35] <cradek> is there going to be a fest this year?
[02:21:35] <jmkasunich> I sure hope so
[02:21:35] <paul_c> 2004 or 2005 ?
[02:21:39] <jmkasunich> 05
[02:21:44] <cradek> next Apr I assume
[02:21:45] <jmkasunich> 04 is just about done
[02:22:16] <asdfqwega> Driving in snow is no fun
[02:22:33] <cradek> jmkasunich: maybe they use a different calendar over there
[02:22:37] <cradek> wasn't it April last year? I drove thousands of miles
[02:22:37] <jmkasunich> yes, april
[02:22:38] <asdfqwega> I'm so far out in the sticks, I'm sure some people 'round here are using the Aztec calender
[02:22:38] <jmkasunich> You're in Va, or Md?
[02:22:48] <cradek> me? NE
[02:22:54] <asdfqwega> OH
[02:23:02] <jmkasunich> I thought you were near Steve Stallings
[02:23:13] <jmkasunich> asdfqwega: me too
[02:23:17] <jmkasunich> Cleveland burbs
[02:23:25] <cradek> I'm not near anything but corn
[02:23:44] <jmkasunich> I must be thinking of somebody else
[02:23:47] <asdfqwega> jmk: were you at the last Fest?
[02:23:55] <jmkasunich> yes
[02:24:01] <asdfqwega> * asdfqwega tries to place faces...
[02:24:02] <jmkasunich> where in OH?
[02:24:15] <asdfqwega> I'm about a couple hours south on I77
[02:24:16] <jmkasunich> beard and glasses (as if that helps in this crowd)
[02:29:18] <asdfqwega> Near New Phila/Dover
[02:29:19] <jmkasunich> hmmm... want to consider sharing some driving?
[02:29:19] <asdfqwega> No problem...if you can stand to be seen in a Volvo
[02:29:19] <cradek> what state is Fest in? I went through so many
[02:29:19] <jmkasunich> can you stand to be seen in a pickup?
[02:29:19] <asdfqwega> I suppose I could stoop that low :P
[02:29:19] <jmkasunich> I'm not sure where Roland's place is - west of NAMES tho, ILL I think
[02:29:19] <paul_c> cradek: The choices are MI, WI, CA, or GB
[02:29:19] <jmkasunich> ??
[02:29:19] <cradek> haha, never heard of GB
[02:29:19] <asdfqwega> GB? That not in Ohio ;)
[02:29:19] <paul_c> with Senator Blair....
[02:29:19] <cradek> I vote for chicago
[02:29:19] <jmkasunich> where did WI and CA come from .... is CA Sherline?
[02:29:19] <cradek> yep
[02:29:19] <jmkasunich> somebody also mentioned (not entirely in jest) Las Vegas
[02:29:19] <paul_c> * paul_c was thinking of a cabin up in Mineral Kings
[02:29:19] <cradek> jmkasunich: it's sure cheap to stay there, I hear
[02:29:19] <jmkasunich> cheap airfare and I guess fairly cheap lodging
[02:29:19] <asdfqwega> paul:Mineral Kings?
[02:29:19] <paul_c> Sequoia NP
[02:29:20] <jmkasunich> betcha there is no HS inet there
[02:29:25] <jepler> I'd have to think pretty hard before going to Vegas
[02:29:28] <asdfqwega> My goodness, I haven't seen a Sequoia in ages
[02:29:39] <jmkasunich> besides, if we were there, I for one wouldn't get any work done - spend all my time outdoors
[02:29:40] <cradek> jepler: me too
[02:29:54] <cradek> jepler: that's a nasty place
[02:30:01] <jmkasunich> I don't particularly like the vegas idea either
[02:30:04] <cradek> jepler: and, too far to drive
[02:30:05] <paul_c> Cable available in Three Rivers just down the road...
[02:31:18] <asdfqwega> jmk: You gave the HAL presentation?
[02:31:26] <jmkasunich> such as it was
[02:31:33] <jmkasunich> not much of a public speaker
[02:31:45] <jmkasunich> the HAL_Intro document is a better version
[02:31:56] <cradek> jepler: seems like we should present AXIS, but what is there to present about it? Lookie here, you can navigate with the mouse. Whee, whee!
[02:32:06] <jmkasunich> (oh, you're matching faces... yes, that was me)
[02:32:37] <paul_c> * paul_c tries to match a face to asdfqwega
[02:32:49] <asdfqwega> cradek: Isn't that the best kind of presentation? Oooo, shiny! <clicky click>
[02:32:58] <cradek> asdfqwega: I could do that kind...
[02:33:10] <paul_c> Yay... PowerPoint
[02:33:23] <jepler> cradek: yeah, and I can drone on about how it was designed and bore everyone
[02:34:05] <jepler> and then we could take questions, like "but AXIS runs like **** on my pentium 120"
[02:34:17] <jmkasunich> that would be me
[02:34:24] <asdfqwega> jepler: Or you could go the perilous route and let people play with it and try to break it
[02:34:33] <jepler> asdfqwega: noooooo
[02:34:36] <jepler> it's my baby
[02:34:42] <cradek> jepler: then I would say "That's not a question"
[02:34:59] <asdfqwega> Statement, 1-luv
[02:35:20] <asdfqwega> "Game of questions" <smirk>
[02:35:41] <jmkasunich> paul_c: what happens if you edit a bash script while it's running?
[02:35:46] <jepler> that was so fun .. back when i was in highschool
[02:36:01] <paul_c> jmkasunich: Not a lot...
[02:36:31] <jmkasunich> do the changes get executed, or does bash read the whole script into memory at the start and then ignore the file while it runs?
[02:36:48] <cradek> jmkasunich: in my experience, it can cause wrong things
[02:37:14] <jmkasunich> * jmkasunich waits for the script to finish first
[02:39:00] <asdfqwega> jmk: In my experience with Mandrake, things are usually read into memory - I can do stuff like delete an MP3 while it's playing. I've even edited Python scripts while running...but I didn't save until I was sure it was finished.
[02:39:23] <cradek> deleting a file only unlinks it to the directory
[02:39:24] <jepler> deleting a file won't cause problems for something slowly reading the file
[02:39:30] <cradek> a program can still have it open
[02:39:32] <jepler> Python reads the whole .py file and compiles it before it does anything
[02:39:46] <jepler> the shell seems to read the file as it goes (I'm not sure how loops are handled, does it seek?)
[02:40:00] <cradek> yeah, neither of those examples are at all like bash
[02:40:53] <cradek> asdfqwega: programs regularly unlink things like temp files long before closing them
[02:42:01] <jmkasunich> moot point now - script finished, edits made, and kicked off again
[02:42:09] <asdfqwega> Hm, learn something every day
[02:42:16] <cradek> jmkasunich: but can't we still argue??
[02:42:20] <cradek> jmkasunich: it's ... what we do
[02:42:21] <jmkasunich> it probably won't fail again for days, but this time I'll have a log of what happened
[02:42:31] <jmkasunich> ;-)
[02:43:26] <paul_c> Time for bed - Catch you guys later.
[02:43:31] <cradek> g'night
[02:43:32] <jmkasunich> goodnight Paul
[02:43:56] <cradek> now what did I come down here to do? I was probably going to do something to axis
[02:48:43] <asdfqwega> * asdfqwega plays "Alzheimer's", looking for his t-square
[02:49:06] <cradek> I like that game too
[02:49:10] <jmkasunich> my father-in-law calls it "Old-Timers"
[02:49:11] <cradek> ... apparently
[02:49:23] <jmkasunich> I call it CRS
[02:49:33] <asdfqwega> Can't Remember Shit
[02:49:38] <jmkasunich> ;-)
[02:52:21] <CIA-9> 03jmkasunich * 10emc2/src/emc/motion/control.c: turned off some debugging stuff to avoid cluttering kernel logs
[03:02:17] <asdfqwega> I just love sitting back with a soda and watching EMC run the machine - it scratches the Mad Scientist Itch
[03:02:32] <jmkasunich> grin
[03:02:35] <asdfqwega> "It's alive! It's aliiiiiiive!"
[03:03:15] <asdfqwega> Damn, I gotta make up a gallery of some of my latest stuff
[03:03:29] <CIA-9> 03cradek * 10emc2/src/emc/task/emctaskmain.cc:
[03:03:29] <CIA-9> fix a manually-applied patch copied from emc1 (a block was deleted
[03:03:29] <CIA-9> when it should have been moved)
[03:03:38] <asdfqwega> Can anybody recommend a good webhost w/ e-commerce?
[03:04:02] <jmkasunich> that fix has been a real bear to get applied ;-)
[03:04:20] <jmkasunich> I think that's the third attempt
[03:04:27] <cradek> I should have done it instead of paul having to figure it out
[03:04:44] <cradek> it's four actually, he's done two and I've fixed both of them
[03:05:02] <cradek> that file is a terrible mess with all its switch/case statements
[03:05:20] <cradek> it borders on the absurd.
[03:05:22] <jmkasunich> I stay away from that stuff
[03:05:35] <cradek> very wise
[03:05:47] <jmkasunich> * jmkasunich prefers kernel modules
[03:05:48] <cradek> (but I wanted that bug fixed and everyone was staying away from it)
[03:06:13] <jmkasunich> I'm glad you fixed it - I wouldn't even know where to start
[03:06:43] <cradek> thanks (but I wasn't fishing for a compliment, I swear)
[03:07:29] <jmkasunich> you've been doing *nix for a long time, right?
[03:08:12] <asdfqwega> Half a cheer for cradek: Hip! Hip! Hoo-
[03:08:14] <cradek> professionally for about 9 years, amateurally (look, a new word) for more
[03:08:40] <jmkasunich> can I brainstorm something with you - has to do with "generally expected conventions" for command line progs
[03:08:47] <cradek> sure
[03:09:10] <jmkasunich> halcmd can accept commands either on the command line (one command) or from a file (can be many commands)
[03:09:32] <jmkasunich> right now, if you want to feed it a file, you say "halcmd -f filename"
[03:09:39] <jmkasunich> if you leave out the filename, it uses stdin
[03:10:00] <jmkasunich> if you say "halcmd foobar", it expects that foobar is a command, not a file
[03:10:22] <cradek> so far, that sounds entirely reasonable to me
[03:10:38] <jmkasunich> if you want to pipe something to it, you have to do "| halcmd -f", not just "| halcmd" like you would with many other progs
[03:11:21] <cradek> with no args, I think it should do the -f behavior
[03:11:44] <jmkasunich> no args = read from stdin, OK
[03:12:07] <jmkasunich> right now, no args gives you a help screen
[03:12:19] <jmkasunich> as does -h
[03:12:21] <cradek> that should happen with -? or -h or --help
[03:12:48] <jmkasunich> I didn't implement -- style long options
[03:12:58] <cradek> ok
[03:13:01] <jmkasunich> perhaps I should ... there is a library for that isn't there?
[03:13:18] <cradek> a special getopt
[03:13:33] <cradek> getopt_long
[03:13:38] <jmkasunich> I seem to recall looking at that, it had more dependencies than I wanted to deal with
[03:13:43] <jmkasunich> anyway, I digress
[03:14:03] <cradek> yeah nobody will complain as long as you have -h/-? document the (short) options
[03:14:26] <jmkasunich> right now, halcmd always returns success (0), regardless of whether the command(s) actually succeeded... I want to fix that
[03:14:51] <cradek> sounds like a good idea
[03:15:09] <jmkasunich> default behavior will be to exit(1) or exit(-1) if a hal command fails, immediately (don't execute remaining commands in the file, if running from a file
[03:15:48] <jmkasunich> but want an option that will tell it to ignore failed commands and keep reading the file
[03:16:00] <cradek> typical exit failure is 1, or use exit(EXIT_FAILURE)
[03:16:14] <jmkasunich> thanks
[03:16:23] <cradek> -k might be a good choice for that, because it's what make uses
[03:17:02] <jmkasunich> that's the kind of info I was looking for... I was thinking -f, as in rm -f, but that conflicts with the -f <filename>
[03:17:28] <cradek> yeah, I think -k is the obvious choice
[03:17:34] <cradek> you can document it as "keep going"
[03:17:47] <jmkasunich> ok, then I don't have to mess with -f
[03:17:56] <cradek> from the make man page: -k Continue as much as possible after an error.
[03:18:21] <cradek> -f is more like "try harder" which isn't really what you are doing
[03:18:27] <jmkasunich> ok
[03:18:39] <jmkasunich> some other options that may or may not match conventions:
[03:18:41] <cradek> (this is all a matter of opinion, of course)
[03:18:56] <jmkasunich> -q quiet, display errors only (the default)
[03:19:02] <jmkasunich> -Q very quiet, display nothing
[03:19:18] <jmkasunich> -v verbose display results of every command
[03:19:34] <jmkasunich> -V very verbose, display lots of junk as things get done
[03:20:14] <jmkasunich> (all documented under -h, but do they meet the "principle of least astonishment"?
[03:20:35] <cradek> sure, they look fine
[03:21:41] <jmkasunich> ok, thanks
[03:21:47] <cradek> happy to help
[03:27:58] <cradek> your loop doesn't look too complex yet, but you should really try getopt if you add more complex stuff
[03:28:09] <cradek> it's easy to use and gets everything right
[03:28:16] <jmkasunich> the only one I'll be adding is -k
[03:28:34] <jmkasunich> what is involved in using getopt?
[03:28:41] <jmkasunich> * jmkasunich hates dependencies
[03:29:52] <cradek> you just call it in a while(option = getopt(argc, argv, "acceptedflags") != -1) {switch(option) {...}
[03:29:57] <cradek> }
[03:30:15] <cradek> acceptedflags will be "qQvVf:k" or something like that for you
[03:30:34] <cradek> the : means it (the f option) takes an argument
[03:31:10] <jmkasunich> what dependencies are involved? if it needs packages not already on the BDIs, I'm inclined to avoid it
[03:31:22] <cradek> no, it's part of the standard library
[03:31:31] <cradek> it's very very old and standard
[03:31:51] <jmkasunich> * jmkasunich does "man getopt" and is enlightened ;-)
[03:31:55] <cradek> #include <unistd.h>
[03:32:13] <jmkasunich> I like "very very old and standard" ;-)
[03:32:26] <cradek> me too
[03:32:54] <jmkasunich> is getopt-long also very old and standard?
[03:33:00] <cradek> not as old, it's a GNUism
[03:33:18] <cradek> but you could call it "old" with a straight face
[03:33:52] <cradek> anyway, just a suggestion, I'm not saying you should rewrite it
[03:34:13] <jmkasunich> but getopt-long is not part of the std lib?
[03:34:24] <cradek> yes, it is part of glibc
[03:34:27] <jmkasunich> if it cleans up the code, maybe I will
[03:34:38] <jmkasunich> but not tonight ;-)
[03:34:40] <cradek> it will always be there on every linux system
[03:35:21] <jmkasunich> can't find a man page for it here...
[03:35:29] <cradek> getopt_long
[03:35:39] <cradek> _ not -
[03:35:54] <jmkasunich> apropos getopt_long -> nothing appropriate
[03:35:59] <cradek> huh
[03:36:09] <jmkasunich> duh
[03:36:16] <jmkasunich> it's listed on the getopt man page
[03:36:23] <jmkasunich> * jmkasunich is blind
[03:36:33] <cradek> funny, my system lets me man getopt_long too
[03:37:16] <jmkasunich> man getopt_long works, but apropos does not, and I apparently typo'ed my first attempt with man
[03:37:38] <cradek> beware the many traps
[03:37:50] <cradek> typos are but one trap that can thwart us
[03:38:02] <jmkasunich> I've been trapped many times
[03:38:06] <jmkasunich> and still I find more
[03:39:17] <jmkasunich> I think I will add long options, eventually
[03:39:26] <jmkasunich> but for today, -k
[03:40:17] <cradek> I'd put it in the tracker and hope someone else would do it...
[03:40:22] <cradek> but that's just me :-)
[03:40:53] <jmkasunich> nah, HAL is my baby...
[04:08:59] <CIA-9> 03jmkasunich * 10emc2/src/hal/utils/halcmd.c: fixed halcmd so that it stops on a failed command, and returns success or failure as appropriate, this can be overridden by specifying -k
[04:29:23] <CIA-9> 03jmkasunich * 10emc2/scripts/emc.run: modified run script to clean up and exit if HAL configuration fails
[05:02:43] <CIA-9> 03jmkasunich * 10emc2/docs/man/man1/halcmd.1: updated halcmd man page to reflect changes made over the last few months
[05:02:55] <jmkasunich> time for sleep
[07:26:37] <TomP> hello, emc user here
[11:25:36] <alex_joni> hello paul_c
[11:29:23] <paul_c> Morning Alex
[11:31:16] <alex_joni> it should be midday.. but it feels like early morning
[11:31:20] <alex_joni> crappy weather here
[11:33:08] <alex_joni> * alex_joni wonders if paul has a flashplayer in his browser
[11:34:25] <paul_c> Eeuwww.... Why would I want a flash player ??
[11:36:15] <alex_joni> to try something ;)
[11:38:19] <paul_c> lemme fire up the laptop and see if that has flash.
[11:40:40] <alex_joni> http://www.transience.com.au/pearl.html
[11:40:49] <alex_joni> it's a nice waste of time
[11:45:08] <paul_c> A game that you can never win
[11:45:51] <alex_joni> but you can
[11:45:52] <alex_joni> ;)
[11:45:56] <alex_joni> I already did
[11:46:19] <alex_joni> when you do.. there is http://www.transience.com.au/pearl2.html
[11:46:24] <alex_joni> that's tougher
[12:04:19] <paul_c> Uggh... Someone wants me to put java on the next Live CD...
[12:06:52] <CIA-9> 03Zathras 07BDI build system * 10Babylon Cluster/0-new.ngc: File changed. New revision:heart.dnc
[12:19:11] <CIA-9> 03Zathras 07BDI build system * 10Babylon Cluster/emctaskmain.cc: File changed. New revision:
[12:26:51] <CIA-9> 03Zathras 07BDI build system * 10Babylon Cluster/changelog: File changed. New revision:conffiles
[12:28:18] <CIA-9> 03Zathras 07BDI build system * 10Babylon Cluster/postinst: File changed. New revision:postrm
[12:38:14] <alex_joni> java?
[12:45:40] <paul_c> didn't like to say it was a hideous waste of space for the Sun SDK...
[12:45:55] <paul_c> but I did ;)
[12:49:13] <alex_joni> lol.. how about apt-get java
[12:53:01] <paul_c> E: Couldn't find package java
[13:05:28] <alex_joni> no java on debian?
[13:07:16] <alex_joni> there is a tool to build debs from REs or JDKs (http://packages.debian.org/unstable/misc/java-package)
[13:07:31] <paul_c> java-common & a java virtual.
[13:08:13] <alex_joni> hope you don't feel offended when I say: I personally don't like Java ;)
[13:09:28] <alex_joni> this is one link about java I kinda like: http://www.democratandchronicle.com/apps/pbcs.dll/article?AID=/20041002/BUSINESS/410020333/1001
[13:10:15] <paul_c> You can dig up as many links as you like.... Java is NOT going to be included.
[13:11:57] <alex_joni> this is actually about a ruling (Kodak vs. Sun) which says: "The jury decided in Rochester on Friday that Sun infringed on technology belonging to Kodak when it developed and introduced Java more than a decade ago."
[13:12:31] <alex_joni> and yes, I don't expect Java on BDI
[13:23:10] <jepler> morning folks
[13:23:15] <alex_joni> morning jepler
[13:24:06] <jepler> what makes you think any language is immune from patent lawsuits? Java is sued because it has a single company behind it which still has deep pockets.
[13:31:16] <jepler> For instance, I bet you can dig up patents on the C99 "restrict" keyword
[19:21:23] <jepler> the place sure isn't busy today
[19:29:32] <paul_c> * paul_c is running tkemc and axis with the same freqmod module.
[19:33:14] <jepler> yeah, I run good old xemc when testing axis, because it's much faster to just restart axis than to start emc again
[19:33:22] <jepler> they get along with each other .. mostly
[19:37:51] <paul_c> Hmm... cds.ngc is, eerrr...
[19:38:04] <paul_c> interesting in preview..
[19:38:27] <cradek> some arcs specified with R don't work
[19:38:56] <cradek> fwiw, it's documented in README
[19:39:28] <paul_c> now who reads the documentation ;)
[19:39:48] <cradek> obviously one of us doesn't
[19:39:51] <cradek> :-P
[19:40:09] <paul_c> Oooo... That was a nasty slap ;)
[19:40:33] <cradek> not if the little guy sticks his tongue out
[19:43:16] <cradek> offset coordinate systems are broken too
[19:43:25] <cradek> maybe other things
[19:43:38] <cradek> but it will all be fixed in 1.0
[19:43:42] <paul_c> Wonder what would be involved in adding some python-vtk routines to do some solid rendering
[19:44:44] <cradek> it wouldn't be a quick hack, to say the least
[19:45:44] <jepler> oh, and canned cycles besides g81 and g82
[19:45:56] <paul_c> well... The new CD has all the vtk stuff on it should anyone want to play.
[20:10:00] <jepler> can I download it yet?
[20:13:03] <cradek> paul_c: is sherline still going to host it with mike gone?
[20:21:24] <paul_c> cradek: As far as I'm aware, Sherline will still host the images.
[20:22:18] <cradek> paul_c: that's good
[20:23:10] <paul_c> I have a couple of other sites that have offered space should we need it.
[20:41:42] <les> ho hum...sick day for me...just playing on the computer
[21:20:17] <paul_c> * paul_c spends way to much tile "playing"
[21:20:43] <les> too sick to do anything else haha
[21:24:49] <les> oh...and perhaps I should rip tile out of the bathroom and sent it
[21:25:24] <les> to you. It is aparently a valuable commodity over there.
[21:26:19] <les> too sick to spell as well.
[22:55:11] <les> Paul, do you know Peter Klosowski? He is your BDI distributor in australia I understand.
[22:55:30] <les> Trying to get him on this IRC.
[22:59:47] <paul_c> * paul_c had been watching CSI...
[23:01:05] <les> I am too sick to watch TV
[23:02:01] <paul_c> does kinda put you of food at times.
[23:02:08] <les> Remember...we have hundreds of channels of garbage
[23:02:18] <les> from orbit...
[23:02:38] <les> oh except BBC....heh
[23:03:55] <les> CSI....Miami? I was planning to spend jan and feb down there
[23:04:17] <les> but that is shot to hell...too much work. Can't take off.
[23:04:32] <paul_c> This was the one from Vegas.
[23:04:38] <les> oh