#emc-devel | Logs for 2006-11-19

[01:41:15] <rayh> Is there an "easy" way to copy all of the wiki contents to my box here?
[01:42:16] <jmkasunich> wget maybe?
[01:42:23] <jmkasunich> not sure how that works on a wiki
[01:42:44] <jmkasunich> SWP could probably pull a copy off the server
[02:04:15] <rayh> I'm getting stuff with wget -xr
[02:06:00] <rayh> Thanks for the hint.
[02:47:39] <jmkasunich> I wonder if there is a way to import man pages into lyx
[02:48:04] <jmkasunich> if a HAL component has a good manpage, I don't want to write the same thing over again for the User Manual
[02:48:15] <rayh> seems like there was a man2latex
[02:48:27] <rayh> let me look.
[02:48:34] <rayh> still getting stuff from wiki.
[02:48:43] <jmkasunich> some things benefit from the Lyx ability to add pictures, etc, but not all need that
[02:50:20] <rayh> mtex2tex Man Page
[02:50:20] <rayh> In the output produced by mtex2tex, the title of the manpage is a LaTeX \section, and the sections of the manpage are LaTeX \subsections. ...
[02:50:42] <rayh> Not certain what that means.
[02:51:48] <jmkasunich> me neither
[02:52:02] <rayh> I'll see if I can find the package.
[02:52:04] <jmkasunich> I guess I'll focus on making good man pages for the moment
[02:52:50] <rayh> I'll look for a converter.
[02:59:23] <rayh> There is quite a bit of interest in converting man to html on the fly for web browsing of these pages.
[03:08:25] <cradek> jepler: in sim/lathe, I'm still sometimes getting spindle override starting at 0%
[03:14:55] <rayh> man2html is a ubuntu package.
[03:46:34] <jmkasunich> cradek or jepler: CVS question related to .cvsignore
[03:46:51] <jmkasunich> the .cvsignore file in docs/man contains a single entry 'man9'
[03:47:00] <jmkasunich> that tells CVS to ignore the entire man9 directory
[03:47:20] <jmkasunich> since I'm adding new manually generated man pages to man9, thats gotta
[03:47:22] <jmkasunich> gotta go
[03:47:46] <jmkasunich> any thoughts on how to replace it?
[03:48:23] <cradek> some of the files in that directory are auto generated, some not?
[03:48:30] <jmkasunich> yeah (now)
[03:48:40] <jmkasunich> before they were all autogen'ed
[03:48:56] <cradek> that's IMO the actual problem
[03:49:04] <jmkasunich> ?
[03:49:31] <cradek> I mean mixing handwritten and auto-gen files
[03:49:53] <jmkasunich> so how can we do this?
[03:50:11] <jmkasunich> stick the handwritten ones in the docs/src tree and copy them on make?
[03:50:36] <cradek> that's not a bad idea
[03:51:25] <jmkasunich> * jmkasunich doesn't want to figure out how to do that ;-)
[03:51:55] <cradek> yeah, can't say I blame you
[03:52:05] <jmkasunich> man1 contains 8 handwritten files
[03:52:19] <cradek> maybe we should wait and see what jepler thinks
[03:52:31] <jmkasunich> man3 contains a crapload that I think were handwritten by jeff (or maybe processed from comments in various .h files)
[03:52:57] <jmkasunich> man9 contains a crapload that are autogenerated from .comp files. plus one (soon more) that was hand written
[03:55:43] <jmkasunich> I wanted to commit the streamer files (one in man1, one in man9) so somebody could look them over and comment before I copy them and revise for sampler
[03:56:19] <jmkasunich> we're not gonna see jeff tonight I don't think
[03:56:23] <cradek> if there are others to be cleaned up, that wouldn't make the problem much worse
[03:56:48] <jmkasunich> well, the one in man9 is the first and only file in man9 as far as CVS knows
[03:56:55] <jmkasunich> it didn't even know about the directory before
[03:57:18] <cradek> hmm ok, hate to add a directory then
[03:57:52] <jmkasunich> welllllll..... seems I already did the cvs add man9
[03:57:57] <cradek> I suspect he'll agree about putting them in docs but I'm not sure
[03:58:08] <cradek> did you commit?
[03:58:12] <jmkasunich> did the cvs add files too, but that doesn't go to the server until I commit
[03:58:14] <cradek> yep I see it
[03:58:19] <jmkasunich> adding a directory takes place immediately
[03:58:22] <cradek> oh
[03:59:58] <jmkasunich> what the heck - I might as well commit the files
[04:00:01] <cradek> crap, I'm out of gcode doubles to specify the in/outfeed angle
[04:00:42] <cradek> I have two letters left, D and L, and they are both ints
[04:01:15] <jmkasunich> integer degrees?
[04:01:36] <jmkasunich> although if somebody is using radians you're screwed
[04:01:36] <cradek> we don't currently use E or N... I'd have to add a new letter and god knows what that takes
[04:02:20] <cradek> the compound angle is read as degrees, can't change it (and it's a double) - 29.5 degrees is commonly used
[04:03:40] <jmkasunich> yeah, for that you definitely need fractional degrees
[04:04:00] <cradek> I guess for the taper it probably doesn't matter so much
[04:04:25] <cradek> I need to have two things: the angle, and whether it's entry or exit (or both??)
[04:04:42] <cradek> if you allow both you probably want to allow different angles, but I don't care about that can of worms
[04:04:54] <jmkasunich> why not have two params, one for entry and one for exit
[04:05:13] <jmkasunich> give both sane defaults
[04:05:18] <cradek> that's a better idea
[04:05:42] <cradek> well the sane default is no taper
[04:05:49] <jmkasunich> 0
[04:07:05] <cradek> I'm very nervous about using up the last two letters of the alphabet
[05:17:09] <cradek> crap N is used, duh
[05:17:12] <cradek> that leaves E
[05:20:09] <cradek> there's a lot ascii available that's not between A and Z... it would sure be a surprise to people though.
[05:20:22] <cradek> a lot of
[05:25:10] <jmkasunich> start using punctuation marks?
[05:25:29] <jmkasunich> I thought you said there were two integers already available?
[05:25:39] <jmkasunich> use them for infeed and outfeed angle
[05:25:47] <cradek> g76 &7 >9.3 <.043 *29
[05:25:59] <cradek> nope, I forgot N is used
[05:26:06] <cradek> (useless, but used)
[05:26:16] <cradek> that leaves only E and some integers
[05:26:40] <jmkasunich> I thought you said there were two INTEGERS already available? ;-)
[05:26:45] <jmkasunich> use integers and be done with it
[05:26:50] <cradek> oh integers
[05:27:04] <jmkasunich> yeah integers
[05:27:09] <cradek> integers?
[05:27:11] <jmkasunich> those county things
[05:30:20] <cradek> maybe you're right
[05:30:25] <jmkasunich> oops: http://www.schlockmercenary.com/
[05:30:38] <cradek> integer degrees makes me sad though
[05:31:43] <cradek> haha
[05:32:11] <cradek> the 'secret life of machines' tried to start a fire with a stick like that, and couldn't make it go even after resorting to a power drill
[05:32:26] <cradek> 'secret life of machines' guy
[05:32:36] <cradek> I keep leaving out words - I should go to bed
[05:32:43] <jmkasunich> it is a bit late
[05:33:03] <jmkasunich> was the secret guy trying to catch the stick on fire?
[05:33:11] <jmkasunich> or just make a hot spot, then add tinder?
[05:33:28] <cradek> not sure
[05:34:45] <jmkasunich> http://mdc.mo.gov/teacher/highered/crafts/craft17.htm
[05:35:07] <cradek> hmm you can only get as many spring passes as you have tools in the tool table. I think that's a bug.
[05:35:24] <jmkasunich> yeah, I think so
[05:36:50] <cradek> Using a bow and drill will not always result in a fire, but it will create a deep affection for matches.
[05:36:58] <jmkasunich> yeah ;-)
[05:37:51] <jmkasunich> in spite of his apparel in this storyline, Kevyn is actually a certified genius and designer of stardrives among other things...
[05:38:35] <jtr_emc> * jtr_emc is back
[05:38:40] <cradek> certified?
[05:38:53] <jmkasunich> 180 IQ or something like that
[05:39:04] <jmkasunich> certifiable sometimes as well...
[05:43:04] <jmkasunich> I have no idea why I'm still awake...
[05:43:06] <jmkasunich> goodnight
[05:43:13] <cradek> night
[06:41:36] <jtr_emc> jtr_emc is now known as jtr_away
[14:04:51] <alex_joni> jepler: nice work on the update script
[14:05:08] <alex_joni> I'm reading through it now.. why are there 3 md5sum's for emc.nml ?
[14:05:46] <jepler> alex_joni: it appears that there were 3 different ones in /etc/emc2/sample-configs
[14:06:06] <alex_joni> yuck :(
[14:06:30] <alex_joni> are you sure?
[14:06:43] <jepler> do you have a machine with 2.0.x installed from .debs? You could check for me
[14:06:48] <alex_joni> halui_halvcp is not part of 2.0.x
[14:06:52] <jepler> md5sum /etc/emc2/sample-configs/*.nml | sort
[14:06:54] <alex_joni> same for ppmc
[14:06:58] <jepler> oh really?
[14:07:19] <jepler> hmm
[14:07:34] <jepler> I guess I checked on a machine where I'd experimented with the 2.1 debs, maybe it's leftovers from that
[14:07:37] <alex_joni> yup.. just one
[14:07:38] <jepler> no ppmc in 2.0?
[14:07:46] <alex_joni> ppmc was removed afaik
[14:07:55] <alex_joni> only univstep and univpwm are left over
[14:08:15] <alex_joni> I can see only one emc.nml in sample-configs for 2.0.4
[14:08:25] <alex_joni> and one additional server.nml and one client.nml
[14:08:31] <jepler> so there is only this one? 57bfc985dae392b1a1d93e7248c259ac
[14:08:38] <alex_joni> that's correct
[14:08:38] <jepler> did any of the sample configs use server.nml or client.nml?
[14:08:47] <alex_joni> no, they are only in common/
[14:09:13] <jepler> so where did this nmlfile come from? 49d422e91c630219b66dd5340708d83d emc.nml
[14:09:20] <alex_joni> http://pastebin.ca/252422
[14:09:28] <alex_joni> jepler: might be pre-2.1
[14:09:29] <jepler> (I see now that this one is the 2.1 one: 4ce3ad0eb076f5ad459f52747dd8fa2a emc.nml)
[14:09:48] <alex_joni> close, but not exactly the same
[14:10:07] <alex_joni> I know it was changed 2-3 times for pre-2.1
[14:10:18] <alex_joni> first when I put emcsvr in charge of the NML's
[14:10:23] <alex_joni> then twice for 64bit stuff
[14:11:57] <alex_joni> the emc.nml inside configs/ppmc is bogus
[14:12:04] <alex_joni> wonder if JonE commited that
[14:12:28] <jepler> etch-servo also has a different nmlfile
[14:12:54] <alex_joni> it seems JonE commited that based on some version he had at the time
[14:13:13] <alex_joni> the one in etch-servo is likewise old
[14:14:00] <alex_joni> I think we should remove them, and get them included by Makefile.. just like for the other configs
[14:14:05] <jepler> yes I'll do that now
[14:14:09] <alex_joni> I can do that too
[14:14:50] <jepler> bbl
[15:01:35] <alex_joni> cradek: when you're around I have a short request
[15:02:28] <alex_joni> there are 2 tcl files in emc2/tcl/scripts (balloon.tcl and emchelp.tcl) which need to lose the executable flag.. maybe you can alter the repo directly to accomplish that.. I found no other way :(
[15:33:58] <jepler> I feel stupid asking this, because it's probably something I checked in by accident, but why is CVS so verbose lately, at least the --enable-sim version?
[15:36:53] <jepler> ah, found it
[15:36:57] <jepler> yay gdb
[15:38:12] <alex_joni> cool
[15:38:15] <alex_joni> what was wrong?
[15:38:27] <jepler> somebody left a line in that was intended for debugging
[15:38:32] <jepler> surprisingly, it wasn't me :-P
[15:40:08] <alex_joni> heh.. ok ;)
[15:40:37] <alex_joni> http://www.labvolt.com/display_left.cfm?view=Topics&ProductID=127&Line_Id=57
[15:40:45] <alex_joni> that's a nice robot to be controlled by emc2
[15:42:33] <skunkworks> lab-volt? why do I cringe when I hear that name?
[15:43:45] <alex_joni> skunkworks: I have nfc ;)
[15:44:04] <skunkworks> I don't know. Odd
[15:44:06] <alex_joni> I don't see it as much more than a toy
[16:36:39] <jmkasunich> jepler: around?
[16:36:51] <jmkasunich> btw, that was me that left the debug line in there, sorry
[16:40:24] <rayh> Hi jmkasunich. I've never left stuff laying around that I didn't really mean to commit.<g>
[16:40:33] <jmkasunich> ;-)
[16:40:55] <rayh> That wget clue worked great.
[16:41:02] <rayh> thanks
[16:41:04] <jmkasunich> cool
[16:44:19] <jepler> jmkasunich: dropping by for a minute .. what's up?
[16:45:38] <jmkasunich> you've used streamer/sampler as much as anybody
[16:45:50] <jmkasunich> could you look at the new manpages and see if I left anything out?
[16:46:07] <jmkasunich> I'm gonna copy the streamer ones as a start on sampler
[16:46:18] <jepler> I'll look
[16:46:22] <jmkasunich> thanks
[16:46:36] <jepler> the other day I had the thought that the userspace part of streamer/sampler might be useful as a library that people can use from their C programs
[16:46:46] <jepler> as opposed to doing popen() and parsing with scanf() or whatever
[16:47:11] <jepler> I didn't look at the code since thinking that, though -- how close/far are they from being a library?
[16:47:42] <jmkasunich> I'm not sure which part you are suggesting they might use, the read file and parse part, or the write to the fifo part?
[16:47:56] <jepler> the read/write FIFO part
[16:48:14] <jmkasunich> ok, that was my first thought, but then you started talking about scanf, etc
[16:48:22] <jmkasunich> lemme look at the code
[16:49:02] <jepler> PINS
[16:49:02] <jepler> streamer.N.pin.M output
[16:49:03] <jmkasunich> there really isn't much to it
[16:49:09] <jepler> is it really pin.m and not pin-m ?
[16:49:28] <jmkasunich> yes
[16:49:31] <jepler> ok
[16:50:25] <jmkasunich> curr_depth should be curr-depth (both the man page and the code use _, so the man page is techically correct...)
[16:50:29] <jmkasunich> I'll fix that right now
[16:50:59] <jepler> it sounds fine to me
[16:51:58] <jmkasunich> re: a streamer lib - streamer_usr.c is just one big main()
[16:52:07] <jmkasunich> (not counting a little signal handler)
[16:52:25] <jmkasunich> about 70 lines of setup
[16:52:56] <jmkasunich> and an 80 line loop that both reads the file and writes the fifo
[16:53:31] <jmkasunich> libifying it would certainly be possible, but wouldn't be trivial
[16:53:58] <jepler> I already forgot the particular use I had in mind
[16:54:39] <jmkasunich> probalby 80% of the code in the loop is the FIFO access, the file I/O is just a call to fgets, and a switch that invokes strtod, strtol, etc
[16:55:37] <jmkasunich> maybe you were thinking of programs that want to examine sampler output for an automated test?
[16:55:52] <jmkasunich> having the numbers in variables saves some parsing
[20:21:15] <alex_joni> anyone around?
[20:21:35] <jepler> alex_joni: off and on
[20:21:48] <alex_joni> * alex_joni wonders if we shall talk about branching today
[20:23:00] <Lerneaen_Hydra> 2.1 already?
[20:23:13] <alex_joni> Lerneaen_Hydra: pre-2.1 ;)
[20:23:19] <jepler> I'm in favor of it, and I think it can be done at any time
[20:23:19] <alex_joni> lots of testing till 2.1
[20:23:38] <alex_joni> jepler: so there's nothing you still favour of including?
[20:24:32] <jepler> I don't want to put it off -- there are so many good things in 2.1 already, that we shouldn't put off branching any more.
[20:24:59] <alex_joni> good
[20:41:58] <Lerneaen_Hydra> so 2.1 testing in a few months or so?
[20:42:37] <alex_joni> 2.1 testign right after the branch
[20:42:46] <alex_joni> release in 1-2 months minimum
[20:44:53] <jepler> the more testers, the less time it will spend in testing
[20:45:51] <alex_joni> jepler: with 2.0.4 all users should have been upgraded to the new repo at linuxcnc.org
[20:46:07] <alex_joni> that means we can use dsplabs as a testing repo, and not harm users
[20:46:18] <Lerneaen_Hydra> oh, so it will be called stable after a month or two?
[20:46:23] <jepler> we'll have to keep that opportunity in mind
[20:46:48] <alex_joni> jepler: sure, for when we'll use the final namings
[21:50:15] <cradek> I'm still in favor of a branch anytime, even though I don't have in the feature I want to do
[21:50:45] <Lerneaen_Hydra> go go css & units/rev! ;)
[21:50:59] <cradek> and that's not it :-)
[21:51:02] <Lerneaen_Hydra> damn
[21:51:04] <Lerneaen_Hydra> :p
[21:51:13] <alex_joni> cradek: how about making a short list
[21:51:35] <cradek> well we keep doing that...
[21:51:44] <alex_joni> even if it involves some extra work, I think branching and allowing the things on that list to be added might be the way to do it
[21:52:45] <cradek> yes we could sure do that
[21:53:23] <alex_joni> at least it's a definate step in the right way
[21:53:38] <Lerneaen_Hydra> is there any feature list that non-devels can see?
[21:53:50] <alex_joni> feature requests & tasks
[21:54:16] <alex_joni> http://sourceforge.net/pm/task.php?group_project_id=46285&group_id=6744&func=browse
[21:54:29] <Lerneaen_Hydra> oh, sourceforge...
[21:54:44] <Lerneaen_Hydra> wasn't that depreciated?
[21:54:51] <cradek> just for cvs
[21:55:14] <Lerneaen_Hydra> oh, right
[21:56:18] <Lerneaen_Hydra> is there a head feature list that non-devels can see?
[21:57:12] <jepler> non-devels see everything devels see
[21:57:37] <Lerneaen_Hydra> oh. ok, is there a head feature list somewhere?
[21:57:46] <jepler> there's this: http://cvs.linuxcnc.org/cvs/emc2/docs/NEWS?rev=HEAD;content-type=text%2Fplain
[21:57:49] <cradek> no secrets!
[21:58:04] <Lerneaen_Hydra> whee OSS
[21:58:12] <jepler> and of course there's all the discussion on this channel and the emc-developers mailing list
[21:58:47] <Lerneaen_Hydra> ok, so no "formal" list as such then :)
[21:59:16] <cradek> does block delete really work now?
[21:59:51] <jepler> I haven't tested it myself
[21:59:53] <alex_joni> cradek: sorta-kinda
[22:00:08] <cradek> ok I'll take both of those as a 'no'
[22:00:15] <alex_joni> it works based on the state at interpretation time
[22:00:29] <alex_joni> it works, but it's not an easy-to-expect behaviour
[22:00:42] <cradek> ok so you have to decide whether to delete before you start the program?
[22:00:58] <alex_joni> yes, it would be best not to touch it during the run
[22:01:06] <alex_joni> that will definately cause confusion :)
[22:01:15] <alex_joni> it somehow works on loops..
[22:01:16] <cradek> ... if anybody uses it
[22:01:17] <jmkasunich> * jmkasunich wonders if this is yet another thing that could be addressed the way we adressed the FO, etc enables?
[22:01:25] <cradek> no
[22:01:30] <jmkasunich> ;-)
[22:01:39] <cradek> you can block delete things other than moves
[22:01:43] <jmkasunich> ok
[22:01:52] <jmkasunich> forget I said anything
[22:01:56] <alex_joni> for instance variable assignment
[22:02:00] <cradek> yeah it has to work at interp time
[22:02:01] <alex_joni> which makes my head hurt
[22:02:06] <cradek> haha
[22:02:37] <cradek> I think you "just" have to not queue past any line that starts a block that might be deleted
[22:02:38] <jepler> if block-deleting a variable assignment makes your head hurt, chew on this: / O100 endsub
[22:02:52] <alex_joni> eek
[22:03:07] <alex_joni> especially if the next one is a / O101 sub
[22:04:22] <cradek> having trouble caring about this...
[22:04:29] <cradek> I don't recall a user ever asking for it
[22:04:37] <jmkasunich> already stopped caring about it ;-)
[22:04:52] <alex_joni> there was a feature request once, that's why I added it
[22:04:58] <cradek> ah ok
[22:05:04] <jmkasunich> actually thats not strictly true - to stop caring, I would have had to cared at some prior time
[22:05:05] <alex_joni> but I'm _very_ ok, if it's hidden somewhere :D
[22:05:53] <jmkasunich> how bout closing the feature request: "will not do" and explaining the horrible nasty ambiguities that can arise?
[22:06:13] <Lerneaen_Hydra> * Lerneaen_Hydra seems to have missed the entire block delete debacle
[22:06:39] <cradek> not a debacle, I think it's just 80% finished
[22:07:30] <cradek> NEWS is not very complete - for instance no mention of jogwheels
[22:07:39] <cradek> or lathe
[22:08:03] <jmkasunich> I'll look back and see when jogwheels were added, and put an entry in NEWS
[22:08:31] <cradek> Lerneaen_Hydra: can you explain 'Lead In - Specifies the tool lead in distance as the number of Pitch lengths the tool must lead in by along the line selected as the drive line for the cycle.'
[22:08:44] <Lerneaen_Hydra> haha
[22:09:42] <cradek> I'm trying to figure out whether that has something to do with a synchronized entrance angle
[22:10:05] <Lerneaen_Hydra> basically it's an acceleration distance, specified in X number of pitch units.. err.. if pitch is 3 and lead in is 2 then the lead in would be 6 in real units
[22:10:10] <cradek> I was thinking about specifying entrance angle as degrees, but number of pitches is another way
[22:10:22] <Lerneaen_Hydra> angle is good to have in degrees
[22:10:39] <cradek> so the tool moves past the ends of the drive line?
[22:10:38] <Lerneaen_Hydra> its just the distance that is typically done in X number of pitches
[22:11:25] <cradek> ok I'm not going to do leadin/out as the wiki says then - you can just change the drive line
[22:11:29] <Lerneaen_Hydra> yeah, so if the thread is to start at Z0 and go towards Z-, pitch is set to 3 and lead in is set to 4 then the thread should start at Z+12
[22:11:58] <cradek> ok
[22:12:05] <cradek> that seems silly :-)
[22:12:26] <Lerneaen_Hydra> it can be good imo if you want to sync where the start of the thread is relative to the rest of the peice
[22:12:49] <cradek> true but it only saves you a tiny bit of math
[22:12:57] <cradek> and I don't have the letters to spare!
[22:12:59] <Lerneaen_Hydra> cnc programmers are lazy :p
[22:13:09] <Lerneaen_Hydra> at least I am
[22:13:32] <Lerneaen_Hydra> how can you do angled lead-ins without it?
[22:13:52] <cradek> well that's what I'm struggling with
[22:13:56] <Lerneaen_Hydra> ah
[22:14:06] <cradek> I was thinking they should be within the drive line, with the angle specified
[22:14:23] <cradek> I don't really like the idea of moving past the drive line
[22:14:26] <Lerneaen_Hydra> then the programmers have to know trig... :p
[22:14:49] <Lerneaen_Hydra> it should be very visible in the axis plot
[22:14:54] <alex_joni> what's a drive line?
[22:14:58] <cradek> that would be a reason to specify it in pitches, not angles
[22:15:16] <cradek> alex_joni: look at g76.ngc in cvs
[22:15:22] <Lerneaen_Hydra> fanuc manages to not run out of chars by dividing it into subrows
[22:15:45] <cradek> yeah they're the one with P12345678 with 123 being one parameter and 45 being another etc?
[22:15:58] <cradek> IMO that's crazy, but I see why they did it
[22:16:01] <Lerneaen_Hydra> so g76(?).1 first bit, and then a g76(?).2 second bit
[22:16:13] <Lerneaen_Hydra> that's the tool setting data IIRC
[22:16:38] <Lerneaen_Hydra> where the letters in .1 and .2 mean different things (which could be an issue if mixed up)
[22:16:51] <cradek> alex_joni: drive line is the overall motion from start to end, and is outside the work (along Z normally), the passes are offsets (in X) from it
[22:17:15] <alex_joni> ahh.. ok, thanks
[22:17:26] <Lerneaen_Hydra> if done well however mixing from 1. to .2 and vice versa should give illegal/bogus/way off values that are plainly visible
[22:17:44] <cradek> I'd rather not do that ...
[22:17:57] <Lerneaen_Hydra> pick your poison...
[22:18:08] <cradek> I have one letter left (E) which could be a double, two ints left but I forget what they are
[22:18:34] <cradek> and I only want one more feature (angled leadin/out)
[22:19:12] <cradek> I could use the two ints for # of leadin/out pitches
[22:19:17] <cradek> that avoids the trig problem
[22:19:55] <cradek> and it's very easy to see if it's a degenerate cut (len/p) < in pitches + out pitches
[22:23:00] <cradek> does the silence mean everyone else thinks that's a stupid idea?
[22:23:08] <Lerneaen_Hydra> yeah! :/
[22:23:35] <alex_joni> the fish doesn't think.. because the fish knows.. everything
[22:23:57] <Lerneaen_Hydra> hmm, I'm still trying to figure out if you can manage to get all the features without being doing bad stuff (r)
[22:24:22] <cradek> Lerneaen_Hydra: have you used what works so far?
[22:24:39] <Lerneaen_Hydra> I've been using single-row g33's and CAM
[22:24:56] <cradek> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/nc_files/g76.ngc?rev=1.2
[22:26:06] <Lerneaen_Hydra> shouldn't that be any number 1<r<2 ?
[22:26:23] <Lerneaen_Hydra> less/greater or equal
[22:26:42] <cradek> no, you can use >2 if you want to be more conservative than constant area
[22:26:57] <cradek> but it should say >=
[22:27:03] <Lerneaen_Hydra> so the area lessens for every cut?
[22:27:07] <cradek> yes
[22:27:31] <cradek> you might want that if you have some deflection
[22:27:32] <Lerneaen_Hydra> what is the function, will inputting very high values break stuff?
[22:27:47] <Lerneaen_Hydra> like r=1e10 or something insane
[22:27:50] <cradek> it'll never really get anywhere...
[22:28:02] <Lerneaen_Hydra> ok, so like e^x then
[22:28:05] <cradek> I'm not sure what will break if you use insane values
[22:28:15] <Lerneaen_Hydra> if x-> -inf
[22:28:29] <Lerneaen_Hydra> what is q?
[22:29:01] <cradek> the angle setting of the virtual compound slide
[22:29:05] <Lerneaen_Hydra> oh!
[22:29:18] <Lerneaen_Hydra> aka exit angle?
[22:29:31] <cradek> no
[22:29:44] <Lerneaen_Hydra> oh, so you mean for tapers?
[22:29:52] <cradek> still no
[22:29:59] <Lerneaen_Hydra> x.x
[22:30:04] <cradek> it lets you cut mostly on one side of the tool
[22:30:24] <Lerneaen_Hydra> oh, aka "flankinmatning" in borkese
[22:30:27] <cradek> look at it in axis and you'll understand
[22:30:38] <cradek> if you say so :-)
[22:30:45] <Lerneaen_Hydra> * Lerneaen_Hydra doesn't have an emc box nearby
[22:31:01] <Lerneaen_Hydra> * Lerneaen_Hydra is waiting for a non-rt sim environment :p
[22:31:01] <cradek> the multiple passes are offset at that angle
[22:31:25] <Lerneaen_Hydra> one lathe I worked with had alternating angles
[22:31:26] <cradek> so at q30 it would cut only on the left side of the 60 tool
[22:31:45] <Lerneaen_Hydra> so every other cut would go on the opposite side for a more even wear
[22:31:49] <cradek> that's a neat idea I hadn't considered
[22:32:03] <Lerneaen_Hydra> what about q>30 for a 60 degree tool?
[22:32:10] <Lerneaen_Hydra> breakage?
[22:32:20] <Lerneaen_Hydra> or just a broken tool
[22:32:41] <cradek> just a bad cut I think
[22:32:49] <Lerneaen_Hydra> ok
[22:32:50] <cradek> can't quite picture what that would do
[22:32:56] <Lerneaen_Hydra> i[.2-.125] ?
[22:32:59] <cradek> it's not what you want :-)
[22:33:06] <Lerneaen_Hydra> definetly not
[22:33:32] <cradek> drive line is at .2, first pass at .125 rad
[22:34:26] <Lerneaen_Hydra> oh, so that defines both the OD and ID of the thread?
[22:35:03] <cradek> drive line radius - i,k are first,last cuts
[22:36:00] <Lerneaen_Hydra> wouldn't defining i,j,k, and r be overdefining the expression?
[22:36:23] <cradek> no, j is the first depth increment only
[22:37:37] <Lerneaen_Hydra> oh, I misread
[22:39:40] <Lerneaen_Hydra> hmm, all those values seem to be needed to make it work
[22:39:53] <Lerneaen_Hydra> and there's still more :/
[22:40:06] <Lerneaen_Hydra> lead in/out angle at least
[22:40:17] <Lerneaen_Hydra> prefferably a lead in/out distance
[22:41:38] <cradek> I think angle and distance is overspecified?
[22:41:49] <Lerneaen_Hydra> huh?
[22:42:12] <Lerneaen_Hydra> how can it be overspecified?
[22:42:32] <cradek> * cradek grumbles about not being able to draw a picture
[22:42:49] <Lerneaen_Hydra> if you know the start point of the "real" thread, you need to know either the start X,Z points of the lead in or the length and angle
[22:43:46] <cradek> if you know the angle why do you need a length too?
[22:44:21] <Lerneaen_Hydra> to know how long the lead in is going to be?
[22:44:37] <jmkasunich> I'm suffering from lack of drawing too, but don't you need to specify both the Z and X distance for the leadin?
[22:45:03] <cradek> you have X distance - it's the last cut minus the first cut
[22:45:11] <cradek> I think the only other thing you need is an angle
[22:45:14] <cradek> that gives you the Z distance
[22:45:27] <Lerneaen_Hydra> oh, we're not talking extending this outside of the OD?
[22:45:32] <jmkasunich> or you specify the Z distance and let it calculate the angle
[22:45:37] <jmkasunich> I dunno which is nicer
[22:45:49] <cradek> jmkasunich: yes that's why I was thinking it could be in 'pitches'
[22:46:24] <jmkasunich> iow "pull out in 1.5 revs"
[22:46:38] <Lerneaen_Hydra> or pull out at 30 degrees
[22:46:41] <cradek> yes or 'my taper should be 6 threads long'
[22:46:43] <Lerneaen_Hydra> both sound nice
[22:47:09] <jmkasunich> I favor "my taper should be 6 threads long"
[22:47:19] <Lerneaen_Hydra> still, being able to extend the pullout beyong the outer diameter wouldn't be bad
[22:47:24] <jmkasunich> closer to reality - few people will actually want a taper of X degrees
[22:47:59] <Lerneaen_Hydra> afaik in most documentation it's defined as an angle, but it's no major thing
[22:48:23] <cradek> I'm starting to think pitches is better than angles
[22:50:20] <cradek> now all I have to do is decide that integer pitches is good enough and I'm set
[22:51:06] <cradek> I do agree pullout past the OD is good
[22:51:13] <Lerneaen_Hydra> as long as the start point can be defined at non-integer points all is good
[22:51:28] <cradek> there's nothing to say the first (I) pass has to be exactly the OD
[22:51:34] <Lerneaen_Hydra> also a tapered start down could be nice if the part is "backwards"
[22:51:46] <Lerneaen_Hydra> but then you've got lots of cuts in the air
[22:51:51] <jmkasunich> integer pitches isn't good enough
[22:52:18] <cradek> jmkasunich: why?
[22:52:42] <jmkasunich> does your large lathe have a threaded spindle nose?
[22:52:58] <Lerneaen_Hydra> if it's in a tight space and you want to taper out to a certain specific point
[22:53:01] <jmkasunich> if you look at it, I bet the thread ends in less than one turn
[22:53:17] <jmkasunich> 0.5, 0.75, 1, 1.5, those are all reasonable values
[22:53:47] <cradek> jmkasunich: you'd just use rapid out then
[22:54:12] <cradek> as I see it, this is just for tapers that you want on purpose
[22:54:29] <jmkasunich> how repeatable is rapid out?
[22:54:36] <cradek> quite
[22:54:40] <Lerneaen_Hydra> uh, is there anything else you were wondering today?
[22:54:48] <jmkasunich> you still need to maintain chip load uniform while you are pulling out
[22:55:42] <cradek> Lerneaen_Hydra: no, thanks, you leaving?
[22:55:52] <Lerneaen_Hydra> for today yes :)
[22:55:56] <jmkasunich> if you are stuck with integers, the I vote for angle, not picth
[22:55:58] <jmkasunich> pitch
[22:56:02] <cradek> Lerneaen_Hydra: goodnight
[22:56:06] <Lerneaen_Hydra> I'll be here tomorrow too
[22:56:10] <jmkasunich> integer pitch is just too coarse IMO
[22:56:16] <Lerneaen_Hydra> * Lerneaen_Hydra agrees with jmk
[22:56:30] <jmkasunich> a 6 pitch taper would be rare, IMO
[22:56:31] <Lerneaen_Hydra> 1 degree isn't much, 1 pitch is a lot
[22:56:42] <Lerneaen_Hydra> anyway, 'night
[22:56:48] <jmkasunich> goodnight
[23:03:58] <alex_joni> I'm off for today..
[23:04:06] <alex_joni> but you have my blessings on a branch :=)
[23:04:07] <cradek> goodnight
[23:04:10] <cradek> ok
[23:04:13] <jmkasunich> goodnight
[23:04:19] <alex_joni> good night all