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

[03:34:57] <jepler> man I really need to figure out why latex2html craps all over some of those very nice images in the documentation
[03:35:18] <jepler> maybe that should go on the task manager...
[03:41:50] <jepler> https://sourceforge.net/pm/task.php?func=detailtask&project_task_id=133070&group_id=6744&group_project_id=46285
[03:49:13] <cradek> latex2html --dont-crap-it-all-up
[13:42:51] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[14:50:59] <alindeman> [Global Notice] Hi all. You may have noticed that services is currently offline. We're working to restore it quickly; however, if its downtime necessitates assistance, please feel free to contact me or any other on-duty staffer ( /stats p ). Thanks again.
[15:01:30] <jepler> I requested that gmane pick up the emc lists, and it appears they have begun to archive the posts. http://news.gmane.org/gmane.linux.distributions.emc.user
[15:01:57] <jepler> they have a process for incorporating old archives in mbox format; we should figure out who has the most complete one and go through that
[15:05:16] <cradek> that sounds good but emc isn't a linux distribution
[15:05:33] <jepler> that's not the group name I requested
[15:05:46] <jepler> * jepler shrugs
[15:05:51] <cradek> I wonder if there's any chance of fixing it
[15:05:59] <jepler> I don't know
[15:44:53] <alindeman> [Global Notice] Hi all! Services should be back up and running. If you notice any issues, please contact an on-duty staffer ( /stats p )
[16:18:09] <alex_joni> hi all
[16:19:38] <cradek> morning alex
[16:19:45] <alex_joni> hey chris
[17:55:05] <Lerneaen_Hydra_> Lerneaen_Hydra_ is now known as Lerneaen_Hydra
[19:52:00] <SWPadnos_> SWPadnos_ is now known as SWPLinux
[20:38:41] <SWPLinux> hmmm. so, readline is a little bit random sometimes
[20:39:51] <SWPLinux> I noticed in halcmd that just pressing return will re-use the last command
[20:40:18] <SWPLinux> which of course isn;t so great if you just press return after loading halcmd, since there is no last command
[20:41:58] <cradek> bug!
[20:42:09] <SWPLinux> yeah - though my initial analysis isn't correct
[20:42:32] <SWPLinux> I fixed the problem for the initial <enter> case
[20:43:01] <SWPLinux> readline always returns an initialized pointer, so you get a non-null pointer to a null string if there's no input
[20:43:24] <SWPLinux> is it a feature that the previous line gets re-executed if you press enter?
[20:44:20] <cradek> "depends"
[20:44:20] <SWPLinux> heh
[20:44:31] <SWPLinux> useful for "show pin yaddayadda"
[20:44:37] <cradek> right
[20:44:41] <SWPLinux> not so useful for "loadrt threads ..."
[20:45:02] <SWPLinux> I guess the qquestion is, should I try to get rid of that feature? ;)
[20:45:18] <cradek> good question
[20:46:18] <cradek> I think the feature is worth the inconvenience(?) of not being able to hit enter willy-nilly
[20:46:39] <cradek> I'm trying to think of what it hurts to leave it
[20:46:52] <SWPLinux> ok. I can check in the fix for the "first line" problem
[20:47:15] <cradek> yeah definitely fix that
[20:47:54] <cradek> when the cat walked between me and the screen, I clicked the window to try to raise it above the cat
[20:48:04] <SWPLinux> did it work? ;)
[20:48:07] <cradek> no
[20:48:24] <SWPLinux> luckily, I'm wearing cat-colored pants right now
[20:48:43] <SWPLinux> they always end up that way, I guess
[20:49:12] <cradek> this cat has both white and black fur, to make that impossible
[20:49:32] <SWPLinux> heh
[20:51:17] <SWPLinux> ok, now this is weird
[20:51:33] <SWPLinux> I haev a fixed version of halcmd, but if I remove the debugging prints, it's no longer fixed
[20:52:05] <SWPLinux> literally, add 2 sets of // to printf lines, and it's back to being buggy
[20:52:43] <cradek> yuck
[20:52:55] <SWPLinux> yeah
[20:54:22] <SWPLinux> nice bug list in the readline man page:
[20:54:30] <SWPLinux> "It’s too big and too slow."
[21:18:03] <SWPLinux> hmmm. well, it seems I was able to fix both problems at once, reagardless of whether I want to or not
[21:21:08] <SWPLinux> ok - it looks like that's the correct fix, since he old behavior isn't really to repeat the last line
[21:21:58] <SWPLinux> it actually only uses the first word on the line, and uses all blank parameters (so "show pin blend" becomes "show", and "loadrt foo" becomes "loadrt")
[21:22:23] <cradek> hmm, that's not very useful
[21:22:27] <SWPLinux> nope
[21:22:34] <cradek> gdb repeats the whole line
[21:23:26] <SWPLinux> the real problem here is in the tokenizer, I think
[21:24:03] <SWPLinux> the line doesn't get reused, but some of the tokensend up with the previous values
[21:36:18] <cradek> arrgh, I don't know how much time I've spent tracking this "interp bug" only to find it's in AXIS somewhere
[21:36:22] <SWPLinux> heh
[21:36:31] <SWPLinux> which one is that?
[21:36:48] <cradek> the first motion after an O code has the wrong sequence (line) number
[21:37:01] <SWPLinux> ah
[21:37:16] <SWPLinux> aren't sequence numbers independent of line numbers?
[21:37:30] <cradek> I don't think so...?
[21:37:41] <SWPLinux> ie, a loop will have increasing sequence numbers, but still execute the same lines
[21:37:58] <cradek> no, the sequence numbers loop too
[21:38:08] <SWPLinux> hmmm - is that the right thing to do?
[21:38:22] <cradek> depends what a sequence number is
[21:38:35] <SWPLinux> I thought the sequence number was an NML thing, for identifying the command for status/response reasons
[21:38:41] <cradek> emc uses them to say 'the motion right now is because of line XXX'
[21:38:47] <SWPLinux> ah
[21:38:53] <cradek> no, that might be a different thing with the same name
[21:38:58] <cradek> serial number maybe?
[21:39:12] <SWPLinux> ok - it's probably my assumption, not any documentation or anything
[21:40:14] <SWPLinux> argh. pointers(/strings) are totally confused in halcmd
[21:40:28] <SWPLinux> I don't think I want to fix it right now
[22:04:30] <rayh> you guys are starting to scare me now.
[22:05:45] <jmkasunich> why's that rayh?
[22:06:01] <cradek> because normally we look like we know what we're doing?
[22:06:53] <rayh> Nah. I know you guys know what you're doing.
[22:07:17] <rayh> I'm really impressed with the recent work. Great stuff.
[22:07:17] <cradek> oh god, we've still got you tricked then
[22:07:36] <rayh> Yep and I plan to remain that way.
[22:09:35] <rayh> btw I talked with alex_joni a bit about a qt4 lib.
[22:09:57] <rayh> Got a guy in the philippines(sp) working on one.
[22:10:34] <rayh> If there is any interest here I believe we could get it lgpl and into the repository.
[22:11:31] <cradek> a lib of what?
[22:12:27] <rayh> emc related commands.
[22:13:03] <rayh> built into qt's signals and slots sort of thing.
[22:24:23] <cradek> is it a NML<->qt interface of some kind? like emcsh?
[22:25:00] <alex_joni> hi all
[22:25:04] <alex_joni> * alex_joni just came in
[22:25:08] <cradek> hi
[22:25:28] <rayh> Yes.
[22:26:54] <alex_joni> finished reading back
[22:42:16] <jmkasunich> the gangs all here
[22:45:52] <alex_joni> seems so
[22:51:39] <alex_joni> rayh: so the question is: do we want the NML<->qt lib in emc2 under LGPL
[22:52:52] <alex_joni> or leave it to the guys who developed it (still LGPL because they need to use the emc.hh)
[22:53:57] <rayh> That's about it.
[22:54:24] <alex_joni> well .. it can be usefull at some point I guess :)
[22:55:21] <alex_joni> how bad would it be if we do it after 2.1 ?
[22:55:31] <rayh> It could also be thought of as cluttering up the repository by requiring one more dev.
[22:55:51] <alex_joni> it's not the cluttering.. but it adds to the dependencies
[22:56:14] <alex_joni> and it will be a bit of work to get it right (configure & co to search for Qt)
[22:56:33] <rayh> IMO the machines using it will be out about the same time or soon after as 2.1 is released.
[22:56:53] <alex_joni> so if it's going in it should be before 2.1
[22:56:55] <rayh> sure.
[22:57:11] <rayh> Well it's not necessary to include with emc2.
[22:58:06] <alex_joni> I understand
[22:59:48] <SWPLinux> I wonder if a gnome-related lib would be more appropriate, considering that ubuntu is gnome-based
[22:59:58] <SWPLinux> not that we can decide for them what they want to do :)
[23:00:12] <alex_joni> you mean gtk ?
[23:00:29] <SWPLinux> yeah, kind of
[23:00:52] <SWPLinux> more like "gtk controls that know what to do with EMC/NML"
[23:01:53] <cradek> like swp I'm a little sad to see reliance on another new toolkit (lerman's doing this too)
[23:02:22] <cradek> like my dad making fun of my mom for buying all the different ziploc bags: "I think there's a size you don't have yet"
[23:02:29] <alex_joni> cradek: we can do this using suggests
[23:02:39] <alex_joni> so only people who install it can use it
[23:03:18] <cradek> that's a packaging issue, I thought we were talking about putting it in our source tree (and compiling it)
[23:03:17] <rayh> IMO there are significant advantages to the qt toolkit.
[23:03:33] <rayh> That is why I suggested it to the customer.
[23:03:43] <SWPLinux> there are some, but aI don't know wnough about gtx+-2.xx to know if there are equivalents
[23:03:53] <rayh> It's not a question of the size of the bag it is the double lock zipper that makes the difference.
[23:03:58] <SWPLinux> argh - ignore typos, please ;)
[23:04:11] <rayh> yea me to, what he said.
[23:04:46] <cradek> would this require I install a bunch of qt/kde stuff to build cvs emc?
[23:04:55] <alex_joni> cradek: not if we mark it optional
[23:04:59] <cradek> or do modern systems have all of it already?
[23:05:09] <cradek> alex_joni: you mean with configure or something?
[23:05:16] <alex_joni> and not build it if the required isnt present
[23:05:15] <jmkasunich> yeah
[23:05:17] <alex_joni> yes, configure
[23:05:22] <jmkasunich> we _can
[23:05:25] <cradek> ok I see
[23:05:27] <SWPLinux> if you intsall even one KDE program (like Kate, in my case), you have pretty much everything you need
[23:05:33] <jmkasunich> _can't_ demand that people have all that stuff though
[23:05:48] <alex_joni> jmkasunich: I totally agree
[23:05:57] <jmkasunich> not to compile something that only 10% of less of users will even want
[23:05:55] <rayh> I agree with jmk on that
[23:06:08] <alex_joni> we don't require Gtk now.. and it's very usefull for halmeter & scope
[23:06:12] <alex_joni> and CL
[23:06:31] <cradek> ok I see how it's the same as the scheme we already have
[23:06:40] <jmkasunich> has anyone verified that you get a usable EMC without gtk?
[23:06:53] <jmkasunich> (obviously no halmeter, scope, or CL, but can you use the rest?)
[23:06:54] <alex_joni> I did at some point
[23:07:02] <SWPLinux> one problem with qt is that any apps created with that toolkit won't have the same look as the other GUI emc apps
[23:07:02] <alex_joni> but itmight hav emorphed a lot since then ;)
[23:07:29] <alex_joni> SWPLinux: show me two emc apps that look the same :D
[23:07:37] <SWPLinux> heh
[23:07:47] <SWPLinux> halmeter and halscope have the same widget sets
[23:07:49] <rayh> We have some issues with different look and feel between the guis now.
[23:07:55] <alex_joni> SWPLinux: still.. worlds appart
[23:08:00] <cradek> I don't think that matters really
[23:08:10] <alex_joni> the only one I can think of is xemc & tkemc ;)
[23:08:20] <alex_joni> still different toolkits
[23:08:20] <SWPLinux> and backplot - don't forget backplot
[23:09:02] <alex_joni> anyways.. that's a sideissue
[23:09:28] <cradek> I think my opinion is that if we can help/coerce/nudge a vendor into making part of their software free, we should, because partially free is better than not free.
[23:09:32] <rayh> I'm pretty certain we don't want to buy into qt the way kde has.
[23:09:58] <SWPLinux> probably not
[23:10:15] <rayh> Although a customer might want it all to be qt.
[23:10:21] <SWPLinux> alex - was it you who was collaborating on the qt toolkit with Paul and me?
[23:10:22] <cradek> I also want those vendors' customers to be able to eschew the non-free part of the software as much as possible (for instance, they might decide to use AXIS)
[23:11:24] <alex_joni> SWPLinux: nope
[23:11:26] <rayh> That would not happen.
[23:11:53] <SWPLinux> ok - must have been anon or some other "a" person :)
[23:12:31] <rayh> Certainly a sherline customer can use axis. I believe quite a few do.
[23:12:45] <SWPLinux> I have the basis for some qt controls that know what to do with NML - such as a position readout that attaches to an axis
[23:13:12] <SWPLinux> I was working on an axis control, with jog buttons, readout, homing stuff, etc.
[23:13:25] <cradek> I meant AXIS as just an example.
[23:13:46] <rayh> Is it a bit like the old vaporgui that I put up in the dropbox -- back when we had one.
[23:13:51] <SWPLinux> but working with Qt was a pain, and it wasn't clear how to make the components usable in a user-designed layout (using Qt Designer or KDevelop)
[23:13:56] <rayh> Sure I understand, Chris.
[23:14:11] <SWPLinux> oh yeah - alex - should we restore all those dropbox items? :)
[23:14:22] <rayh> The qtdesigner is what the current library is aimed at.
[23:14:39] <alex_joni> SWPLinux: working on it already
[23:14:38] <rayh> * rayh answers for alex -- no
[23:14:43] <SWPLinux> ok
[23:14:47] <SWPLinux> uh - withe rway :)
[23:14:51] <alex_joni> rayh: I have it locally
[23:14:54] <SWPLinux> either way
[23:15:05] <cradek> I don't know what vaporgui is/was
[23:15:15] <alex_joni> a misty interface
[23:15:18] <alex_joni> * alex_joni guesses
[23:15:30] <SWPLinux> vapor.tgz, 75059 bytes, March 15, 2005 :)
[23:15:36] <rayh> Back in the day when paul was threatening qt I wrote a display using designer.
[23:15:54] <rayh> I put a screenshot on the dropbox.
[23:15:57] <SWPLinux> ok - that's around the time we were working on the NML-aware controls
[23:16:06] <rayh> Right.
[23:16:31] <rayh> That is a nice concept. NML-aware controls.
[23:16:46] <rayh> Or a nice way to describe such things.
[23:16:51] <SWPLinux> yeah - partly for the GUI designer thing that Robin keeps harping on
[23:16:52] <alex_joni> rayh: I see a vapor.tgz is that it?
[23:18:04] <rayh> Seems like that is what I called it.
[23:18:07] <SWPLinux> wow - what's in the 32M file "mazak-atc.tar"?
[23:18:36] <SWPLinux> Alex: I can transfer those directly to DH if you like, unless you're doing that already
[23:18:43] <rayh> beats me.
[23:19:01] <alex_joni> SWPLinux: all kinds of information about mazak
[23:19:03] <SWPLinux> that's big enough for a full kernel, or several copies of the emc source tree
[23:19:09] <jmkasunich> mazak-atc.tar contains a crapload of pics that tomp took
[23:19:09] <alex_joni> wires & pinouts & schematics and all
[23:19:11] <SWPLinux> ok - pdf or image files and the like
[23:19:12] <SWPLinux> got it
[23:19:34] <alex_joni> SWPLinux: I copied most of the stuff, left out some big files
[23:19:51] <alex_joni> in 5 minutes maybe you can put the rest (I assume your connection is faster)
[23:19:58] <SWPLinux> ok. I'm not worried about space or bandwidth. if you think it may be useful to have them there, stick them on
[23:20:12] <SWPLinux> I'd just wget from a DH login
[23:20:25] <SWPLinux> the dropbox is available on metalwroking.net again
[23:20:27] <alex_joni> heh.. good idea.. forgot the old page is online again
[23:20:30] <SWPLinux> working ...
[23:20:32] <SWPLinux> heh
[23:20:56] <alex_joni> you found the dir I made?
[23:21:05] <SWPLinux> interestingly, linuxcnc.org is up to ~15G/day of transfer
[23:21:20] <jmkasunich> are there live-CDs in that mix?
[23:21:21] <alex_joni> that's quite a bit
[23:21:29] <alex_joni> only one ;)
[23:21:31] <SWPLinux> plus ~3.5G/day at cncgear, which has some BDI ISOs still
[23:21:48] <alex_joni> seems like quite a ration (15 vs. 3.5)
[23:22:09] <SWPLinux> it's only BDI on cncgear, and I think there are other mirrors
[23:22:18] <jmkasunich> the bdis at cncgear are rather old aren't they?
[23:22:27] <SWPLinux> no, I think 4.50 or 4.52 is on there
[23:22:32] <jmkasunich> oh
[23:23:21] <SWPLinux> yes
[23:24:02] <alex_joni> ;-)
[23:25:16] <alex_joni> http://www.linuxcnc.org/dropbox/emc-man3.png
[23:25:19] <alex_joni> rayh: was that it?
[23:28:58] <SWPLinux> ok - it looks like the transfer from metalworking will be done in a minute or two
[23:29:52] <SWPLinux> unless his server chokes first ;)
[23:42:40] <alex_joni> well guys.. if there's nothing else.. I'll head for bed
[23:42:50] <alex_joni> getting late, and I'm travelling tomorrow to a customer
[23:42:55] <SWPLinux> nighty-night alex