#emc-devel | Logs for 2008-12-01

[00:12:00] <jmkasunich> jepler: I have a list (in python), whose members are strings
[00:12:04] <jmkasunich> some may be duplicates
[00:12:09] <jmkasunich> I want to remove the dups
[00:12:22] <jmkasunich> in shell, I'd pipe thru sort | uniq
[00:12:36] <jmkasunich> py has sort, but I don't see the equivalent of uniq
[00:12:40] <jmkasunich> any hints?
[00:13:48] <jmkasunich> looks like "sets" might be relevant
[00:15:32] <BigJohnT> you could use a try to see if you all ready had the entry
[00:15:40] <BigJohnT> try:
[00:15:46] <BigJohnT> for x in s;
[00:15:58] <BigJohnT> s:
[00:17:40] <BigJohnT> or brute force
[00:17:41] <BigJohnT> u = []
[00:17:43] <BigJohnT> for x in s:
[00:17:45] <BigJohnT> if x not in u:
[00:17:47] <BigJohnT> u append(x)
[00:17:48] <BigJohnT> return u
[00:18:08] <jmkasunich> it looks like "sets" do exactly what I need, and are available in 2.4 and later
[00:18:23] <BigJohnT> ok, cool
[00:20:13] <BigJohnT> I see that now, sets is cool
[00:20:39] <jmkasunich> I didn't know you did python BigJohnT
[00:20:55] <jmkasunich> now that I do, prepare to be asked questions
[00:21:19] <BigJohnT> yea, I did most of the g code generators on the wiki site in python
[00:21:42] <BigJohnT> ok, but be prepared for c++ questions :)
[00:21:57] <jmkasunich> C questions maybe
[00:22:00] <jmkasunich> I don't do C++
[00:22:04] <jmkasunich> it gives me a rash
[00:22:19] <BigJohnT> ok LOL
[00:22:29] <jepler> jmkasunich: list(set(l)) will uniqify the list, but discard the order
[00:22:42] <jepler> (convert l to a set and back)
[00:22:58] <jmkasunich> jepler: I dropped the list completely, I just create an empty set and add items to it as I find them
[00:23:04] <jmkasunich> if they are dups, set ignores them
[00:23:20] <jepler> yeah -- working in sets in the first place may make more sense
[00:23:35] <jepler> when it comes time to output them, you can " ".join() them just like lists
[00:23:53] <jmkasunich> right now I'm checking for four cases
[00:24:00] <jmkasunich> 1) this file defines entity foo
[00:24:04] <jmkasunich> 2) this file needs entity foo
[00:24:11] <jmkasunich> 3) this file defines package bar
[00:24:17] <jmkasunich> 4) this file needs package bar
[00:24:23] <jmkasunich> (for all files in the directory)
[00:24:58] <jmkasunich> I wind up with two global dicts, one that has entity:file-that-defines-it pairs, the other with the same for packages
[00:25:21] <jmkasunich> and I wind up with two sets per file, with the entities and packages that the file needs
[00:25:48] <jmkasunich> well, not quite yet
[00:26:06] <jmkasunich> at the end of each file, I have a pair of sets, which get erased and refilled for the next file
[00:26:15] <jmkasunich> haven't figured out how to save them yet
[00:26:47] <jmkasunich> list of filename:things-it-needs tuples? dict of the same?
[00:27:09] <jepler> or just print each one out as a makefile rule as you determine it?
[00:27:49] <jmkasunich> while reading file A I figure out that it needs entity foo, but I won't know till much later that file Z defines foo
[00:28:01] <jepler> ah
[00:28:23] <jepler> sounds like you'll need to access things by name, not by order, so it sounds like you may want a dict
[00:28:55] <jmkasunich> { file:things-it-needs, next-file:things-it-needs } ?
[00:30:32] <jepler> i think you'll need a second dict for entity name->filename
[00:31:06] <jepler> so then you can look through things-it-needs for the entities, and loook up the filename from the entity name
[00:31:25] <jmkasunich> yeah, thats what I'm doing (I think, just a sec)
[00:31:39] <jepler> maybe you don't ever need to look things up by filename(?). if so, then that doesn't need to be a dict
[00:31:46] <jmkasunich> # these two dicts contain 'item:file-that-defines-item' pairs
[00:31:46] <jmkasunich> entities_defined = dict()
[00:31:46] <jmkasunich> packages_defined = dict()
[00:31:46] <jmkasunich> # these two dicts contain 'file:set-of-things-it-needs' pairs
[00:31:46] <jmkasunich> entities_needed = dict()
[00:31:47] <jmkasunich> packages_needed = dict()
[00:32:50] <jepler> sounds like what I'm thinking
[00:32:54] <jepler> bbl, I should be socializing
[00:32:57] <jmkasunich> ok
[00:47:22] <jmkasunich> hmm, I think I found a problem with the vhdl sources
[00:47:43] <jmkasunich> hostmot2.vhd needs entity WordPR, but none of the .vhd files supplies it
[00:48:23] <jmkasunich> arcg - vhdl is case-insensitive
[00:48:42] <jmkasunich> wordpr.vhd provides wordpr, hostmot2.vhd needs wordPR
[00:48:50] <jmkasunich> I can fix that
[00:59:15] <jmkasunich> hmm, that fixed, I found another one
[00:59:32] <jmkasunich> this one is real - sserialram is used in sserial.vhd, but not defined in any source file
[00:59:45] <jmkasunich> emailed peter about it, maybe it just a module that isn't used yet
[01:00:26] <BigJohnT> did you get the sets working?
[01:00:32] <jmkasunich> yep, worked great
[01:00:40] <jmkasunich> I have all the data parsed, now I'm working on output
[01:00:45] <BigJohnT> cool
[01:01:17] <BigJohnT> I'm playing with Qt and C++
[01:13:34] <CIA-42> EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/ (Submakefile docs.xml index.tmpl): add g code examples to html
[01:18:29] <CIA-42> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (docs.xml index.tmpl): add g code examples to html
[01:18:59] <BigJohnT> * BigJohnT wanders upstairs to relax
[01:19:05] <BigJohnT> goodnight
[02:38:31] <LawrenceG> BigJohnT, hi
[02:38:52] <BigJohnT> hi
[02:40:28] <LawrenceG> a python question.... I would like the code functionality of truetype tracer in a python function..... string in, list of segments out... have been looking at things like pyrex... is there an easy way to embed c in python?
[02:41:20] <jepler> LawrenceG: the major ways are pyrex, ctypes, and writing extensions directly in C/C++
[02:41:26] <jepler> they all take some learning
[02:42:03] <jepler> personally, in the past I favored writing the extension directly, but both pyrex and ctypes are quite neat in that they let you do more in pure Python or in a language that is strongly reminiscent of Python
[02:43:27] <LawrenceG> jepler, I noticed :}... I am a little concerned about the complexity of the truetype lib headers
[02:44:35] <jepler> LawrenceG: yeah -- tt is a bit daunting at first (but if you've read ttt.c then you know it's tractable)
[02:45:14] <LawrenceG> I could try and redo ttt in python.... in there a simple python way to call tt libs?
[02:45:31] <LawrenceG> is there a ...
[02:46:22] <LawrenceG> things in ttt like the rendering callbacks seem to be one of the major complications
[02:46:25] <jepler> I've looked in the past for a pre-wrapped version of the truetype lib but haven't found it
[02:48:17] <LawrenceG> there was a PyFT project around 2000 but it was a wrapper using fretype 1 interface... there is mention of PyFT2, but it doesnt seem to exist
[02:48:26] <jepler> that sounds like what I remember finding
[02:49:16] <LawrenceG> I also found swig, which is supposed to make wrapping things easier
[02:49:42] <jepler> yeah -- there are a lot of solutions out there
[02:49:46] <jepler> everybody likes different ones
[02:50:16] <LawrenceG> but my python is not strong yet... wrapping ft sounds like a lifetime project :}
[02:50:37] <jepler> it sure would be diving in with all feet
[02:51:32] <LawrenceG> how about a python program calling ttt and using the stdout from ttt to fill a list in python?
[02:52:13] <LawrenceG> that sounds like a simpler interface
[02:52:40] <jepler> sure; you could add a new format to ttt that is easy to parse in python, and fits with what you're trying to do ..
[02:54:30] <BigJohnT> LawrenceG: what are you working on?
[02:54:49] <LawrenceG> thanks.... I'll take a look at the python system calls and see what I can come up with... the mods to ttt would be simple
[02:56:11] <LawrenceG> BigJohnT, ways to do vform engraving.... archivist and I are tossing around algorithms
[02:56:41] <BigJohnT> ok, so your looking at ttt for the outline?
[02:57:38] <LawrenceG> yes... just change ttt to give a series of small segments around exterior of font.... use V cutter to "fill" interior
[02:58:04] <LawrenceG> ttt already chops curves into line segments
[02:58:43] <LawrenceG> would do the same to the lines
[02:59:55] <BigJohnT> cool
[03:00:30] <jepler> BigJohnT: by the way, I saw that you added stuff about parport address in the latency test document .. is that what you intended?
[03:00:54] <BigJohnT> yes
[03:01:19] <BigJohnT> it seemed to fit kinda there...
[03:02:22] <jepler> ok, it was a little surprising to me, but whatever
[03:03:04] <BigJohnT> If you think it fits better somewhere else I can move it...
[03:10:34] <BigJohnT> I was kinda thinking that the name of the lyx file was wrong after I added it...
[03:16:16] <jmkasunich> jepler: dump python question - how do I print something without getting a newline (I want to put more output on the same line later)
[03:18:22] <SWPadnos> put a comma at the end of the list of things to print
[03:18:35] <SWPadnos> print "hello", "there", these_vars,
[03:18:36] <jmkasunich> thank you
[03:18:45] <SWPadnos> sure - does it work? :)
[03:18:53] <jmkasunich> yep
[03:19:01] <SWPadnos> cool
[03:19:35] <SWPadnos> oh, apparently there's also a sys.stdout.write() function
[03:20:03] <jmkasunich> heh, I've moved on
[03:20:17] <jmkasunich> it is a complication that vhdl doesn't have anything like .o files
[03:20:18] <SWPadnos> whatever works :)
[03:20:29] <jmkasunich> (one .o file per input .vhd file)
[03:20:36] <SWPadnos> .rtl?
[03:20:50] <SWPadnos> yeah, I guess that's true
[03:21:24] <jmkasunich> the normal way to handle dependencies is to generate foo.o : things.vhd that.vhd foo.vhd needs.vhd
[03:21:39] <SWPadnos> yep
[03:21:58] <jmkasunich> don't have a .o, what am I gonna do?
[03:22:05] <SWPadnos> you had that taken care of in your m5i2x makefile I though
[03:22:31] <jmkasunich> yeah
[03:22:35] <jmkasunich> I made a .t file
[03:22:39] <jmkasunich> dunno what t stood for
[03:22:42] <SWPadnos> heh
[03:22:46] <SWPadnos> temp?
[03:22:51] <SWPadnos> transition?
[03:22:56] <jmkasunich> it simply contained the names of every needed file
[03:23:02] <jmkasunich> tag maybe
[03:23:13] <jmkasunich> it just served as a placeholder of sorts
[03:25:00] <jepler> jmkasunich: or import sys and use sys.stdout.write() which never adds those newlines you don't want
[03:25:19] <jepler> argh SWPadnos already covered that :-P
[03:25:21] <SWPadnos> heh
[03:25:25] <SWPadnos> google FTW! :)
[03:25:37] <jepler> I don't get to show off my extreme intelligents(sic) when I'm not prompt
[03:25:41] <jepler> but at least the cats have been fed
[03:25:48] <jmkasunich> here's another chance
[03:25:57] <jepler> * jepler leaves in a hurry
[03:25:58] <jmkasunich> given foo.vhd, what is the easiest way to get foo
[03:26:08] <SWPadnos> mv foo.vhd foo
[03:26:17] <jmkasunich> in python
[03:26:22] <SWPadnos> oh, hm
[03:26:31] <jepler> >>> import os
[03:26:32] <jepler> >>> os.path.splitext("foo.vhd")[0]
[03:26:32] <jepler> 'foo'
[03:26:56] <jmkasunich> thank you
[03:27:03] <jepler> >>> os.path.splitext("bar/foo.vhd")[0]
[03:27:04] <jepler> 'bar/foo'
[03:27:06] <BigJohnT> rstrip
[03:27:16] <SWPadnos> ok, that was the next question :)
[03:27:21] <jepler> BigJohnT: bzzzt!
[03:27:32] <BigJohnT> yes
[03:27:32] <jepler> >>> "broth.vhd".strip(".vhd") # wtf?
[03:27:33] <jepler> 'brot'
[03:27:51] <SWPadnos> mmm. or brotchen
[03:27:51] <jepler> (same result for rstrip)
[03:28:16] <jmkasunich> * jmkasunich figured out that rstrip would't work
[03:29:27] <cradek> os.system("basename foo.vhd .vhd")
[03:29:27] <jepler> (strip's argument is a group of characters to strip, not a substring)
[03:30:09] <BigJohnT> is the period always the seperator?
[03:30:54] <jepler> BigJohnT: for splitext, yes
[03:31:12] <SWPadnos> yeah, theoretically you could strip everything that isn't a period, then remove the last character (which is theoretically a period now)
[03:31:17] <jmkasunich> jepler: sys.stdout.write only works with strings, right?
[03:31:20] <jepler> welllllll let me hedge just a bit
[03:31:26] <SWPadnos> but if you have a file with no extension, there may be no period, so you;d lose the whole name
[03:31:35] <jepler> jmkasunich: yes, if you have something 'x' that's not a str, you can write(str(x)) or write("%s" % x)
[03:32:02] <jmkasunich> does write take printf style formatting? (%)
[03:32:12] <BigJohnT> I see
[03:32:16] <jepler> BigJohnT: os.path.extsep is the separator for os.path.splitext and related functions
[03:32:20] <jmkasunich> I assumed it only took a single string, like C write just takes raw bytes
[03:32:43] <jepler> jmkasunich: The "%" operator of strings is similar to printf
[03:32:51] <jepler> stuff like %10d or %15.15f works like in C
[03:33:03] <jmkasunich> ok, what you just said caused an ah-ha moment
[03:33:06] <jepler> other stuff is different like the fact that %s takes an object of any type
[03:33:08] <BigJohnT> I thought you were working with strings
[03:33:12] <jmkasunich> I assumed that % was handled by print
[03:33:22] <jmkasunich> it is a string thing, works with _any_ string!?
[03:33:40] <SWPadnos> the ah-ha (which I also had) is that "%s" %x evaluates to a string with %s replaced by x
[03:34:06] <SWPadnos> (and whatever else you can do with format strings and variables
[03:34:08] <SWPadnos> )
[03:34:33] <jmkasunich> * jmkasunich replaces all his prints
[03:34:37] <jepler> jmkasunich: you can use "format" % arg or "format" % (arg1, arg2, ...) almost anywhere
[03:34:47] <jmkasunich> (was gonna have to do that anyway, output goes to a file, not to stdout, when this is done
[03:34:48] <jepler> I say "almost" because I'm hedging again
[03:35:06] <SWPadnos> eval (blah blah) :)
[03:49:20] <BigJohnT> good night all
[04:01:18] <seb_kuzminsky> paul_c wants to run a buildslave, but i'm worried about his motivations
[04:01:53] <SWPadnos> you could mention that to him
[04:04:11] <seb_kuzminsky> i did
[04:04:22] <SWPadnos> ok, that's good :)
[04:05:19] <jepler> how much trust does the buildmaster have to have for a buildslave?
[04:05:48] <seb_kuzminsky> not much
[04:06:14] <seb_kuzminsky> the master tells the slave what to do, and gets back logs of stdout & stderr and an exit code
[04:06:23] <seb_kuzminsky> that's it
[04:06:33] <SWPadnos> the only thing that should happen is that we get more information on where the build may fail
[04:06:55] <seb_kuzminsky> the master produces status reports on the web and irc based on that information, so i guess a slave could cause the master to spam me
[04:07:00] <jepler> in the past, paul has sent messages to the list saying that emc won't build on his system, but not included useful information to allow troubleshooting the problem. My uncharitable interpretation of that was that he was more interested in making us look bad in public than actually help us
[04:07:07] <SWPadnos> the only bad thing that may happen is that he's running a system that we don't support, and that makes our "stats" look bad (and spams IRC with failure messages)
[04:07:11] <seb_kuzminsky> but no one else is on the buildmaster mailing list yet (i haven't announced it yet)
[04:07:45] <SWPadnos> if, however, we get useful logs of the failures, that would still be a good thing
[04:07:54] <seb_kuzminsky> i told him if he adds new OSes, he's signing up to help fix the build problems
[04:08:00] <SWPadnos> heh
[04:08:14] <seb_kuzminsky> we'll see if he'll step up to the plate here (to use one of them "sports" analogies)
[04:08:21] <SWPadnos> that would be good, he's a good programmer
[04:12:47] <seb_kuzminsky> buildmaster: hello
[04:12:47] <buildmaster> yes?
[04:13:21] <jepler> oh yes, it was in this thread, after he wasted at least a day of my time and a day of chris's time trying to actually troubleshoot the alleged problem with the cvs server: http://thread.gmane.org/gmane.linux.distributions.emc.user/7223/focus=7484
[04:16:03] <seb_kuzminsky> that sucks jepler... :-(
[04:16:49] <jepler> it was one of those times when I was thinking, maybe there's really a problem here, maybe he's not just trying to jerk my chain
[04:18:11] <jepler> anyway, goodnight
[04:18:41] <SWPadnos> see you
[04:20:57] <seb_kuzminsky> seeya jeff
[05:13:21] <jmkasunich> well I think I got dependency tracking working again, this time in a way that doesn't require a specific VHDL coding trick
[05:13:39] <jmkasunich> tomorrow: figuring out how to specify the device
[05:15:14] <SWPadnos> cool
[05:19:07] <seb_kuzminsky> nice jmkasunich
[05:25:22] <cradek> argh, both my "really nice" ebay jog wheels have one channel dead
[05:25:37] <SWPadnos> oh, bummer
[05:25:47] <SWPadnos> was that the $200-ish pair?
[05:25:57] <cradek> $100 iirc
[05:26:02] <SWPadnos> oh, good :)
[05:26:12] <cradek> bah
[05:26:23] <SWPadnos> well, not as bad anyway :(
[05:30:11] <cradek> one is currently being sacrificed for science
[05:30:37] <SWPadnos> just hope it's the driver chip instead of the lamp/sensor/wheel
[05:31:07] <cradek> lamp/led and sensor seem to work
[05:31:46] <cradek> the thing is quite macroscopic
[05:31:47] <SWPadnos> well, that's a good thing
[05:31:55] <SWPadnos> big parts?
[05:32:13] <cradek> yeah
[05:32:19] <SWPadnos> also good
[05:36:21] <cradek> not meant to be serviced...
[05:37:01] <SWPadnos> is it more or less a low-res encoder, or some other mechanism?
[05:38:18] <cradek> just a low res encoder
[05:47:58] <cradek> after prying it apart, it works
[05:52:17] <seb_kuzminsky> cradek: that "not meant to be serviced" stuff stinks
[05:52:36] <seb_kuzminsky> glad it decided to start working
[05:52:40] <cradek> it was all glued together, but 30 year old glue breaks
[05:53:05] <cradek> ha, I fixed it
[05:53:10] <cradek> no idea how
[05:53:35] <cradek> I blew out the slots with compressed air - they looked ok, but maybe one was gooped up.
[05:53:40] <SWPadnos> heh
[05:53:42] <SWPadnos> cool
[05:54:17] <seb_kuzminsky> a gooped encoder-disk slot shouldnt cause one channel to fail, right?
[05:54:34] <SWPadnos> unless that was a sensor slot
[05:54:41] <seb_kuzminsky> dont the sensors share slots?
[05:54:43] <cradek> the disk is huge, but there are very fine slits over the sensors
[05:54:55] <cradek> no they are 3/4 inch apart on this - it's huge
[05:55:01] <seb_kuzminsky> whao
[05:58:15] <SWPadnos> 3/4 +/- a little probably :)
[07:42:25] <CIA-42> EMC: 03swpadnos 07TRUNK * 10emc2/src/hal/user_comps/gs2_vfd.c:
[07:42:25] <CIA-42> EMC: Add command-line options for serial parameters, module name, and debugging messages
[07:42:25] <CIA-42> EMC: Also added some usage instructions.
[07:48:39] <alex_joni> heh, someone's up late
[07:49:04] <CIA-42> EMC: 03swpadnos 07TRUNK * 10emc2/src/hal/user_comps/gs2_vfd.c: oops - didn't want those in caps
[07:49:05] <SWPadnos> yeah
[07:49:10] <SWPadnos> that's about to change though :)
[07:49:16] <alex_joni> reading back.. lots of stuff :)
[07:49:37] <SWPadnos> yeah, there was a bit flying by, wasn't there?
[07:52:07] <SWPadnos> well, the compile farm has had enough time to complain. night :)
[08:04:15] <alex_joni> heh
[08:04:21] <alex_joni> compiling isn't everything :P
[08:04:26] <alex_joni> but.. good night
[11:55:15] <micges> what position is returned in s.actual_position and s.position ?
[11:55:36] <micges> s.position is changing and s.actual_position isn't
[12:32:02] <CIA-42> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ladder/classic_ladder.lyx: add info about s32's
[12:50:39] <jepler> micges: a look in axis.py should help you confirm this, but I think they are "commanded" and "actual" positions.
[13:10:41] <CIA-42> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (docs.xml index.tmpl): add driver to html
[14:40:58] <jepler> hi skunkworks!
[14:46:28] <skunkworks> Good morning jepler!
[14:47:03] <jepler> hi skunkworks!
[14:47:08] <jepler> oh, wait, I said that once already
[14:49:17] <skunkworks> Did you have a good thanks giving?
[14:49:21] <cradek> hi skunkworks!
[14:49:30] <jepler> skunkworks: yep, I sure did.
[14:49:32] <skunkworks> Good morning cradek!
[14:49:37] <jepler> dinner with family and friends in town
[14:49:45] <skunkworks> Nice.
[14:50:06] <skunkworks> we had wild turkey. Pretty good.
[14:50:32] <skunkworks> (plus we didn't have to cook it - just everything else :))
[14:51:39] <skunkworks> cradek: how about you?
[14:51:51] <cradek> I had a lot of fun too
[14:52:20] <cradek> I spent most of vacation helping a friend fix up her old truck
[14:52:36] <cradek> I can't relax on vacations, I go crazy
[14:52:44] <skunkworks> heh
[14:52:46] <cradek> also, there were chocolate-covered cranberries
[14:53:01] <skunkworks> ooh - that sounds awesome.
[14:53:21] <cradek> I made two unobtainable truck parts for the repairs - one was a mill job, one was lathe
[14:53:55] <skunkworks> nice - what was the truck?
[14:54:09] <cradek> 81 jeep cherokee
[14:54:24] <cradek> it's really deluxe - power everything, very nice
[14:54:31] <cradek> only problem is it's as old as dirt
[14:56:18] <skunkworks> heh
[14:57:15] <cradek> after we rebuilt the carb we drove around and did various things - every time she started it, she said "that's just amazing".
[14:58:14] <cradek> the main problem was that all the gas would leak out of the bowl, so it would start very hard because it had to pump the bowl full before it would work
[14:58:49] <cradek> also, the accelerator pump was put together wrong - seems like I always find parts in wrong when I rebuild a carb.
[15:02:49] <skunkworks> funny.
[19:26:00] <alex_joni> SWPadnos: isolcpus will be deprecated soon
[19:26:09] <SWPadnos> so I noticed in your email
[19:26:40] <alex_joni> there was some talk on the RTAI mailing list about it a while ago, but Paolo didn't sweat it much :)
[19:27:06] <alex_joni> btw, he also mentions that 2.6.27 and 28 don't work at all with rtai for him
[19:27:12] <SWPadnos> bummer
[19:27:24] <alex_joni> (actually after the ADEOS-IPIPE patch things tend to not work anymore)
[19:28:10] <SWPadnos> incidentally, did you notice that there's an rtai package in Ubuntu (maybe Hardy, but it could have beenIntrepid)
[19:28:32] <jepler> no, I didn't.
[19:28:40] <alex_joni> oh, there always was one in 'contrib' I think
[19:28:59] <alex_joni> even in dapper
[19:29:00] <alex_joni> http://packages.ubuntu.com/dapper/rtai-source
[19:29:36] <alex_joni> http://packages.ubuntu.com/search?keywords=rtai
[19:29:37] <jepler> $ apt-cache showsrc rtai
[19:29:38] <jepler> Package: rtai
[19:29:38] <jepler> Binary: rtai, rtai-source, rtai-doc
[19:29:38] <jepler> Version: 3.2-1.1
[19:29:39] <SWPadnos> well, there's one called "rtai" in intrepid
[19:29:48] <SWPadnos> righto
[19:29:52] <jepler> it's a very old rtai here on hardy
[19:29:56] <alex_joni> 3.2 is really old
[19:30:00] <SWPadnos> it did seem a little small (at 34kb or something) :)
[19:30:09] <alex_joni> it seems to be 3.6.1-1 for intrepid and jaunty
[19:30:28] <alex_joni> SWPadnos: only docs
[19:30:38] <SWPadnos> that was a separate package IIRC
[19:30:40] <alex_joni> but it depends on librtai1 and librtai-dev
[19:30:49] <SWPadnos> ok
[19:30:49] <alex_joni> http://packages.ubuntu.com/intrepid/i386/rtai/filelist
[19:31:13] <alex_joni> librtai1 only contains libxrt.so
[19:31:24] <SWPadnos> one of these days (or those days) I'll look at making an RTAPI-POSIX layer
[19:31:45] <SWPadnos> which would be good for the -ll kernels as well as just about any RT kernel pathces
[19:31:50] <SWPadnos> or should be anyway
[19:31:57] <SWPadnos> patches
[19:32:22] <jepler> SWPadnos: start here: http://axis.unpy.net/01190912545
[19:32:22] <alex_joni> you do that
[19:32:31] <jepler> "The "port" consists of changing sim_rtapi to use a mostly-pthreads API instead of pth, making it setuid, and making it able to do I/O. "
[19:32:57] <SWPadnos> sounds like a good start :)
[19:33:31] <SWPadnos> there are a couple of stepconf changes that may take precedence though (since they're easy :) )
[19:33:45] <SWPadnos> I'll add G540/G251 timings to it, for one
[19:33:51] <jepler> great, I was hoping for a new stepconf maintainer
[19:33:55] <alex_joni> haha
[19:34:00] <SWPadnos> good luck :P
[19:34:06] <alex_joni> someone's screwed :)
[19:34:56] <SWPadnos> I was thinking of putting the timing and I/O presets into a separate text file so they'd be more easily changed by others
[19:35:29] <alex_joni> text file?
[19:35:39] <jepler> I'm not sure how that helps
[19:35:57] <jepler> "get this text file, do something complicated (possibly with sudo) to put it in the right place, and after typing a dozen terminal commands you can avoid typing in 4 numbers"
[19:36:18] <SWPadnos> no, download this text file into your user directory, then run stepconf
[19:36:41] <SWPadnos> stepconf can look for a file of a specific name and add those settings to its list
[19:36:57] <SWPadnos> err, presets
[19:37:20] <alex_joni> *.stepconf_separate_preset_file
[19:37:47] <SWPadnos> *.stepconf-timings *.stepconf-iosetups
[19:37:52] <SWPadnos> whatever
[19:38:04] <alex_joni> see.. you made skunkworks go away
[19:38:20] <SWPadnos> finally
[19:38:20] <alex_joni> and http://www.megasinnlos.de/userbilder/bilder/you-make-bunny-cry.jpg
[19:38:22] <SWPadnos> :)
[19:38:25] <SWPadnos> heh
[19:39:47] <alex_joni> and http://content.pimp-my-profile.com/userpics/funny_pictures/scaredkitty.jpg
[19:40:06] <SWPadnos> or possessed
[19:40:13] <SWPadnos> -s
[19:40:23] <SWPadnos> +s
[19:40:33] <SWPadnos> s/-s//
[19:53:25] <tomaw> [Global Notice] Hi! One of our servers had a short-lived problem causing some users to be disconnected. It should be back online now, sorry for the downtime and thanks for using freenode!
[20:00:15] <alex_joni> jepler: it does load faster
[20:19:32] <jepler> alex_joni: you mean my website?
[20:21:47] <alex_joni> yeah
[20:21:54] <alex_joni> pictured from DH
[20:21:58] <alex_joni> pictures* even
[20:22:49] <jepler> good to get more confirmation
[20:22:58] <jepler> from home it just feels slower :-P
[20:23:08] <alex_joni> hope you didn't sign up for dh
[20:23:16] <alex_joni> do some loadbalancing :)
[20:24:12] <jepler> Ingrid has an account for her organization. I'm piggybacking on that.
[20:24:43] <alex_joni> I can get you on mine too..
[20:24:50] <alex_joni> * alex_joni has unlimited foos
[20:24:59] <jepler> Ingrid only has 350.2GB or something
[20:25:11] <alex_joni> Disk Usage:
[20:25:12] <alex_joni> 0.3 GB of unlimited
[20:25:22] <alex_joni> Bandwidth Usage:
[20:25:23] <alex_joni> 0.2 GB of unlimited.
[20:25:59] <jepler> yeah yeah
[20:30:08] <alex_joni> BigJohnT: around?
[20:30:23] <BigJohnT> pretty much
[20:32:03] <alex_joni> hi :)
[20:32:11] <alex_joni> * alex_joni is looking for something, jas
[20:32:19] <BigJohnT> ok
[20:34:57] <alex_joni> 2.3 is coming up sometimes next year
[20:35:06] <alex_joni> what do you think about publishing docs?
[20:35:18] <alex_joni> (I mean are they in good enough shape?)
[20:35:30] <BigJohnT> cool, next year is one month away
[20:35:40] <alex_joni> yeah, but 12 months long :)
[20:36:16] <BigJohnT> they could some clean up... still looking at the asciidoc, docbook, html, pdf chain
[20:36:45] <BigJohnT> could use some
[20:36:56] <jepler> I don't think we'll be changing the documentation system for 2.3 ..
[20:37:10] <BigJohnT> but for the most part I keep both sets as up to date as possible
[20:37:58] <alex_joni> I was investigating using a service like: http://www.lulu.com/ a while ago
[20:38:03] <alex_joni> and I think it's a good idea
[20:38:20] <jepler> I should also say that the lyx2lyx bug regarding conversion to old versions has been fixed, so we can stay with lyx and still support old systems
[20:38:31] <alex_joni> oh, really?
[20:38:46] <BigJohnT> so, I can do docs in 8.04 now?
[20:39:07] <alex_joni> if you install a new lyx2lyx I think
[20:39:08] <jepler> BigJohnT: no, I have to make some packages and write a hundred lines of makefile first
[20:39:34] <BigJohnT> ok
[20:39:37] <jepler> but for 2.4 I sure intend to make it possible, while not precluding building the docs on 6.06 machines still
[20:39:49] <jepler> for 2.3 I'm undecided about what to do, and for 2.2 I don't intend to try to do this at all
[20:39:51] <alex_joni> BigJohnT: stumbled upon this http://eneas.juve.ro/~juve/emc/emc2_user_manual_cover_v2.png (reminded me..)
[20:40:34] <BigJohnT> cool, was it for a printed manual
[20:40:40] <alex_joni> yeah, paperback
[20:40:48] <jepler> I should probably follow up about this on the mailing list .. I view it as being up to the people who are more active writing documentation, whether to try to stick with lyx or change systems. I'm not writing so much documentation these days..
[20:40:50] <SWPadnos> has anyone found a usable manpage editor? (by usable I mean one that hides all that nroff/groff/troff crap from me)
[20:40:52] <alex_joni> or hardcover with sleeve
[20:41:31] <BigJohnT> the asciidoc chain can do man pages
[20:41:40] <BigJohnT> I haven't tried any yet
[20:42:16] <alex_joni> BigJohnT: think you know the current pdf pagecount (for the user manual..)
[20:42:22] <alex_joni> * alex_joni could look it up
[20:42:35] <BigJohnT> about 150 or so
[20:43:27] <BigJohnT> the one I have here is 149 pages including all the stuff in the back...
[20:43:29] <alex_joni> I'd have guessed more than that
[20:43:44] <alex_joni> 19$ for a 150 hardback book
[20:43:53] <BigJohnT> I've been busy :)
[20:44:12] <alex_joni> in bw (not colour)
[20:44:46] <alex_joni> 9$ for paperback
[20:47:43] <skunkworks> alex said colour
[20:48:16] <skunkworks> thats soooo european
[20:48:28] <jepler> alex_joni: is that for 6x9? the manual is probably rather longer if it's appropriately formatted for 6x8 instead of just scaled from letter/a4
[20:49:24] <jepler> though selecting letter only makes it $10.75 (same for a4)
[20:49:46] <jepler> coil bound, which is probably the best style .. "perfect" bound is a bit less
[20:51:56] <jepler> first price break at qty25, $8.35ea (a4/letter) / 7.00ea (6x9 "trade")
[21:04:41] <alex_joni> well.. personally I'd get a hardcover with casewrap
[21:05:01] <alex_joni> I'm not sure what size I made that, but I can easily make other sizes
[21:05:12] <alex_joni> skunkworks: well excuse me mr. 'mericun
[21:05:13] <BigJohnT> well, I'm off to the house... talk to you later :)
[21:05:20] <alex_joni> later
[21:05:46] <jepler> will casewrap stay open flat
[21:05:46] <jepler> ?
[21:05:51] <alex_joni> err.. I meant hardcover with dust jacket
[21:05:59] <alex_joni> not fully flat
[21:06:04] <alex_joni> like a regular book :)
[21:06:52] <skunkworks> :)
[21:10:48] <jepler> formatting the Master_User.lyx for 6x9 and reducing the margins slightly gives a 226-page document, and many images are improperly sized to the page
[21:12:42] <jepler> the text seems a bit big, but that's hard to judge onscreen. the smallest available point size (10) is chosen in lyx document layout
[21:14:03] <jepler> (for dust jacket, lulu only has 6x9)
[21:16:28] <alex_joni> yeah, that's too bad
[21:25:51] <skunkworks> jmkasunich: this is in the spec for the ir2184 'Logic and power ground +/- 5 V offset'
[21:26:45] <skunkworks> jmkasunich: so maybe having the drivers hooked to motor 'ground' and everything else hooked to the resistor might be a workable solution.
[21:27:00] <skunkworks> (and would tighten things up_
[21:27:01] <skunkworks> )
[23:16:17] <cradek> seb_kuzminsky: does the hm2 encoder support non-x4 mode (one count per full cycle instead of one count per edge)?
[23:18:07] <seb_kuzminsky> cradek: no i dont think so
[23:18:31] <seb_kuzminsky> what is non-x4 mode used for?
[23:20:24] <cradek> jog wheels mostly
[23:20:52] <cradek> they give a full cycle for each click
[23:21:55] <cradek> I'm not smart enough to know whether it could be emulated in the driver
[23:22:11] <cradek> a mod-4 thingy seems easy enough until you consider index
[23:22:38] <seb_kuzminsky> i think so... return "count >> 2" for count? leave index alone?
[23:22:50] <seb_kuzminsky> brb
[23:22:54] <cradek> me too, dinner
[23:27:18] <seb_kuzminsky> hoo boy, paul_c's buildslave can't check out the cvs tree
[23:30:02] <jepler> o. f. f. s.
[23:30:40] <seb_kuzminsky> well it's very complicated you know
[23:31:29] <jepler> I can think of several obvious things; they're all the things that were suggested in that thread I linked to last night where we already played this game once
[23:34:10] <jepler> at present there's one IP address which is blocked at the firewall between cvs and the internet; it's a chinese address that heavily abused the webserver one day, ineptly running some kind of web mirroring program
[23:35:40] <jepler> paul's extremely bright. you should invite him to troubleshoot it.
[23:38:40] <seb_kuzminsky> the buildslave setup procedure i sent him (http://emc2-buildbot.colorado.edu/buildslave-admin-guide.html) says to make sure you can do a "cvs checkout" before talking to me
[23:39:02] <seb_kuzminsky> he did some crazy cp from somewhere else instead of verifying that cvs co worked
[23:39:10] <seb_kuzminsky> i told him to fix it on his end
[23:39:11] <seb_kuzminsky> we'll see
[23:39:54] <jepler> oh and by the way, 'grid display' linked from http://emc2-buildbot.colorado.edu/buildbot/ is a 404
[23:40:24] <seb_kuzminsky> yeah i know, it's a bug in hardy's buildbot
[23:40:33] <seb_kuzminsky> i'll upgrade it one day...
[23:40:35] <jepler> ok then
[23:53:37] <CIA-42> EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/config/stepper.lyx: some clean up
[23:56:31] <CIA-42> EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/ (Submakefile docs.xml index.tmpl): add tweeking steppers to html
[23:57:32] <seb_kuzminsky> tweek tweek
[23:58:30] <BigJohnT> peep peep
[23:58:39] <CIA-42> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (docs.xml index.tmpl): add a section or two to html
[23:59:04] <BigJohnT> how is it going seb_kuzminsky
[23:59:28] <seb_kuzminsky> hi BigJohnT :-)
[23:59:40] <seb_kuzminsky> i'm digging all the work you're doing on the docs, thanks!
[23:59:51] <BigJohnT> cool, thanks