#emc-devel | Logs for 2010-07-07

[02:28:46] <mozmck> jepler: do you think my kernel is good enough for a livecd?
[02:29:11] <mozmck> I keep removing stuff and still have 1.5 meg left to go...
[02:39:51] <jepler> mozmck: I've used it on only one system, where it worked just fine. I don't know enough about the other issues you've encountered to say whether it's "good enough"
[02:40:15] <jepler> remember to include the hostmot2 firmware packages! That'll probably take a meg or two away from you :(
[02:40:44] <mozmck> The only issue I have now is that I can't get it to run on my 6-core, and I think jthornton had the same problem on a quad-core
[02:40:58] <mozmck> oh, yeah, where do I get that package?
[02:44:29] <mozmck> What do you think about replacing openoffice with abiword and gnumeric? That was an easy way to get all the space I needed and a little more! So far I've removed all games, rhythmbox and related, several fonts, f-spot, pitivi, tomboy, evolution and related, all ubuntuone stuff, transmission (bittorrent client), and all the mono and *cil stuff.
[02:47:39] <jepler> personally, I don't care about openoffice, abiword *or* gnumeric
[02:47:53] <jepler> get the hostmot2-firmware packages from the dapper repository; they install on 10.04 just fine
[02:48:07] <jepler> 0.7 is the current version, but I probably have to release 0.7.1 because I screwed up a dependency
[02:59:04] <mozmck> ok. I figured it would be good to have a word processor and spreadsheet anyhow, and abiword and gnumeric can open a lot of formats and are *much* smaller than openoffice.
[02:59:15] <mozmck> is there something you like better than those?
[03:00:38] <jepler> I don't see the use of a lot of word processing on an emc pc that's not net connected (can't install whatever package is desired)
[03:00:52] <jepler> I can see that spreadsheets would sometimes be useful, but I don't know what alternatives there are to gnumeric
[03:02:33] <mozmck> I don't either. For the word processor, my main thought was for opening and reading documents, not so much for writing.
[03:03:27] <mozmck> Lots of various docs not in pdf format out there.
[03:04:25] <jepler> again, I wouldn't know a thing about how well abiword opens doc or docx files
[03:04:46] <jepler> but if you're downloading random docs directly to that machine -> you have net access and can install what you need
[03:05:18] <mozmck> That's true. abiword seems to do fairly well at doc files, I've never had a docx.
[03:06:06] <jepler> it's sleep time here. Thanks again for working on this! We'll have to talk soon about getting a correctly-formatted apt repo working for 10.04
[03:09:35] <mozmck> no problem! wish I'd had time to get it done sooner.
[03:09:52] <mozmck> I think I'll nuke openoffice
[03:13:33] <cradek> that sounds good to me now that I read what you guys just said
[03:16:59] <mozmck> good! about ran out of stuff to get rid of. I think I may put in gnumeric and abiword because there will be plenty of room.
[03:52:33] <mozmck> heh, removing openoffice.org: Need to get 0B of archives. After unpacking 254MB will be freed.
[03:52:40] <cradek> wow
[03:52:56] <cradek> wonder how much difference that will make when squashed
[03:53:56] <mozmck> Not that much, but quite a bit! I think on the other livecd I did with 9.10, I was able to install inkscape, gnumeric, abiword, qcad, and heekscad and maybe something else along with emc2.
[03:54:21] <cradek> have you managed to package heeks?
[03:54:43] <mozmck> no, that was a package that someone else made.
[03:55:00] <mozmck> I think it was linked to in an email or something.
[03:57:13] <mozmck> what do you think about putting inkscape and/or qcad in?
[03:58:10] <cradek> sadly I'm still the wrong one to ask. I don't use either of those but I understand they are both useful for use in combination with emc.
[03:58:10] <cradek> (specifically how, I don't know)
[03:59:21] <mozmck> I think mostly in conjunction with a CAM program, but there is a gcode output plugin available for inkscape. I don't know how well it works.
[16:51:41] <Guest827> thanks!
[17:50:41] <seb_kuzminsky> if someone will make linuxcnc.org/mozmck into an apt-gettable debian archive, i'll make the buildbot build emc2 debs for lucid
[18:03:51] <mozmck> hey seb, shouldn't the archive be linuxcnc.org/lucid or something like that?
[18:06:15] <cradek> yeah it'll get moved to fit in the existing scheme ... somehow
[18:07:34] <SWPadnos> shouldn't it be something like /debs/dists/<dist-name>/<arch/type>?
[18:08:37] <SWPadnos> I think we almost used that scheme with /emc2 instead of /debs, but stopped (or maybe that's what we're doing, and the /hardy dir has older packages or something)
[18:12:38] <seb_kuzminsky> i agree, the lucid packages should be integrated into the current debian archive structure, or both of them should be reshaped to be more consistent (the current one is rooted at /hardy, which would be surprising if there were lucid packages there too)
[18:19:35] <cradek> I find www.linuxcnc.org/dists and www.linuxcnc.org/emc2, neither of which must be right
[18:19:49] <cradek> (because neither has 2.4/hardy packages)
[18:20:23] <cradek> /emc2 is the one I maintained for brezzy+daaper
[18:20:34] <SWPadnos> http://linuxcnc.org/hardy/
[18:20:42] <SWPadnos> I think that's the one for 8.04
[18:21:03] <cradek> you are right
[18:21:07] <SWPadnos> yielding the strange path /hardy/dists/hardy/xxx to actually get packages
[18:21:36] <SWPadnos> oh, did we change the signing key between using /emc2 and /hardy?
[18:22:23] <cradek> yes
[18:28:55] <CIA-2> EMC: 03seb 07master * re64571334a80 10/debian/control.in: fix a debian control file syntax warning in lucid
[18:33:02] <seb_kuzminsky> speaking of the buildbot... are we ready to turn off 2.3 builds? or is there still life in it?
[18:34:46] <cradek> there are a lot of people still using it, but I think chances are a little low that we'd make another release of it. I see there are no changes since 2.3.5 waiting for release
[18:39:03] <seb_kuzminsky> i guess it doesnt cost anything to leave the 2.3 buildbot running
[18:39:20] <seb_kuzminsky> just a bit more junk on the waterfall page, but it's off on the right where it's not in the way so much
[18:44:14] <KimK> And speaking of repository (and related) changes, I see that the "hostmot2-firmware-5i22-1.5" has popped up on my Update Manager, and under the changes tab it says "Changes for the versions:<newline>0.6<newline>0.7<newline><newline>This change is not coming from a source that supports changelogs.<newline>" In case there is any interest in supporting changelogs. Just thought I'd mention it.
[19:07:42] <KimK> Also, this message pops up when installing any of the hostmot2 software offered: ("Warning: You are about to install software that can't be authenticated...", etc.) Picture at: http://imagebin.ca/view/CilxpCD.html Is there anything that can be done about that in the future?
[19:10:30] <KimK> * KimK should probably stop complaining about *free* software, lol
[19:12:26] <Dave911> Has anyone ever tried running an interpreter independant axis by feeding a position into limit3 hal component and then pipe the output of the limit2 to a stepgen for driver control?
[19:15:12] <Dave911> Jepler: CL Modbus works fine connected to a Prosoft Modbus RTU module stuck in an Allen Bradley SLC 500 PLC rack. Error free so far. I wish the module was as easy to configure as EMC2! The manual for that module stinks.
[19:19:26] <jepler> Dave911: thanks for the update.
[19:19:37] <jepler> I shudder to think what your PLC is like if you think setting up CL Modbus is easier :-/
[19:21:34] <Dave911> The people who wrote the Prosoft Modbus manual should be shot. Even with 100+ pages of text, and an example configuration, I couldn't figure it out without calling the company's hotline.
[19:21:47] <jepler> KimK: last time we looked, it wasn't possible for us to supply changelogs in a way that worked with the update manager. If you look into it and find out different, let us know
[19:23:06] <jepler> KimK: as for the "can't be authenticated" message, is that an 8.04 system configured with the linuxcnc packages automatically (e.g., from linuxcnc.org live cd or emc2-install.sh), or from e.g., a 10.04 system?
[19:23:37] <jepler> if the latter, then you probably omitted the step of adding the linuxcnc repository signing key to your system
[19:25:17] <KimK> jepler: This is on 10.04 and mozmck. OK, thanks, I'll make sure I add the key and try again. As it happens, a visitor arrived and I haven't installed anything as of yet, so I can still retry.
[19:26:17] <Dave911> The PLC maker, Allen Bradley/Rockwell, is getting rather nasty on support. At one time they had an open web forum where you could ask questions and look for solutions. Now you have to have a support contract before you can get into the forum so you can find the info to implement their hardware! Complete BS.
[19:26:39] <jepler> KimK: I think that the same incantation used in emc2-install.sh is appropriate to try on 10.04:
[19:26:43] <jepler> gpg --keyserver pgpkeys.mit.edu --recv-key 8F374FEF
[19:26:45] <jepler> gpg -a --export 8F374FEF | sudo apt-key add -
[19:26:49] <Dave911> Oh well.. I've got it working so far....
[19:27:00] <jepler> Dave911: that's good to hear
[19:28:27] <Dave911> Jepler: Did you ever find a reason why CL Modbus will not run properly without the CL GUI running? I haven't tried to figure that one out yet..
[19:28:41] <cradek> Dave911: yes I'm sure people have done that... why not
[19:29:26] <Dave911> cradek: I can't think of a reason... I was just hoping that I was not overlooking something obvious.. I'll set it up for testing..
[19:29:29] <jepler> Dave911: no, I didn't figure it out
[19:30:45] <Dave911> jepler: OK.... I'll try and figure it out after this next startup which begins on Friday and goes on for a week and a half..
[19:31:13] <jepler> Dave911: I mean, we both found serious bugs in it after I tried adding it, but I haven't looked into fixing it..
[19:31:54] <Dave911> Oh ... I thought that it ran ok on your machine but did not on mine..
[19:32:16] <Dave911> It makes me feel better that you found it broken also. :-)
[19:34:42] <Dave911> At first glance it looked like the modbus sequencing was getting all screwed up without the GUI. The trace messages coming up in the terminal window appeared to be out of order. Or at least quite different from what they were when the GUI was running.
[19:45:50] <jepler> Dave911: The problem I saw was that starting without gui and then later loading the gui deleted the whole modbus configuration.
[19:46:00] <jepler> Dave911: I'm not sure whether I saw the problem you describe or not
[19:50:33] <Dave911> The configuration did not disappear for me, although I think things began to get unstable at that point, so that might have been why you saw the configuration vanish. I'll try it again today and see what I find. I'll back up the config first before trying this out!
[20:15:13] <cradek> how on earth do you home http://www.linuxcnc.org/images/fbfiles/images/belt_schematic.jpg
[20:16:59] <andypugh> Does anyone have a link to the pyvcp widget patch that allows you to use the "enter" key to set the value of a spinbox after manually typing in, (rather than requiring a flick-up / flick-down sequence)
[20:17:02] <micges> cool :)
[20:17:45] <andypugh> I think that is how the flat-bed plotters work
[20:44:02] <Dave911> cradek: How do you control that thing?? What kind of kinematics config would that require? That is an awful lot of belt to stretch and spring.
[20:44:31] <cradek> Dave911: it's an idea someone's discussing on the ... list or forum, I forget which
[20:44:55] <cradek> kinematics look simple to me, but I'm not sure how it would home, and I agree it seems floppy
[20:46:10] <Dave911> Has anyone worked out a simple homing routine using only hal components?
[20:46:27] <cradek> you mean for that geometry?
[20:47:13] <Dave911> No sorry.. referring to what I mentioned before about driving an interpreter independant axis via a limit3 and a stepgen...
[20:48:29] <cradek> stepgen can't zero its position, so you'd be stuck with maintaining your own motor offset
[20:49:55] <Dave911> So when Axis (the GUI) homes a stepgen, it is merely maintaining an offset to zero?
[20:50:31] <cradek> AXIS doesn't home, the motion controller does - and yes it keeps track of the offset (in 2.4 you can even see this offset on a hal pin)
[20:51:26] <Dave911> OK ..
[20:53:11] <cradek> jepler: a while back you did the work to make macros in touchy. if you still have that work, would you put it somewhere safe before it's too late? I'm pretty sure you didn't push it to a shared branch.
[20:53:30] <jepler> cradek: I have no idea where that code went
[20:54:10] <cradek> too late :-/
[20:54:37] <cradek> I don't remember whether it was easy or hard. seems like it was somewhere in between.
[20:54:44] <jepler> well maybe I know
[20:54:44] <jepler> jepler@india:~/emc2-dev$ git branch | grep touch
[20:54:45] <jepler> touchy-macro
[20:55:10] <cradek> hey whaddya know
[20:55:45] <jepler> looks like it's all in one commit: http://emergent.unpy.net/files/sandbox/0001-let-O-subs-be-called-from-MDI.patch
[20:56:10] <cradek> thanks, I'll stash it somewhere too
[20:56:12] <jepler> I think we shelved it because we were interested in probing from macros and that was badly busted
[20:56:12] <andypugh> Dave911: Remembering the offset is pretty easy. The mux component makes a handy sample-and-hold.
[20:56:27] <Dave911> It would be easiest if I kept the axes totally independent without the motion controller... due to the way the machine is controlled presently..
[20:56:56] <Dave911> andypugh: That's right... I looked at that hal code you did before I still have that somewhere...
[20:57:03] <andypugh> There is a limit3 function in HAL which can do most of a single-axis motion control task.
[20:57:29] <Dave911> Yep.... I think I got that idea from you. :-)
[20:58:10] <andypugh> Back to the workshop for an hour I think.
[20:58:32] <Dave911> I need to dig up that info again I guess..
[20:59:42] <Dave911> I think you used a DDT into a Mux as I recall. That was clever...
[21:04:34] <CIA-2> EMC: 03seb 07master * r158f777daaba 10/src/hal/drivers/mesa-hostmot2/ (bitfile.c bitfile.h hm2_7i43.c): complete the "linux firmware is const" conversion