Hi mshaver, Thanks for the response
BTW, why not start out in your endeavor with the high quality, well support Python code in AXIS?
I'd like to be able to use QT designer as I am not a TK/TCL guy but I'm catching onto Python. I'd use Glade but it is sort of clunky IMO. I'm not trying to make an Axis look a like. I work on custom machines mostly that aren't at all machine tools. Mostly purpose specific machines etc.
your best bet, assuming that you want to avoid GPL issues (for whatever reason) is to use something like emcrsh to do the dirty work
It would be valuable to be able to see a good example of how to tie in a QT application to EMC2. That is pretty much it.
QT designer "looks" really nice..
your code, since it would communicate over a socket, would not be required (by emc2 licensing requirements) to be GPL2
if you use Qt free, even to start writing a program, the resulting program must be GPL (v2 I think)
that's what's in the link mshaver pointed out
I am really not concerned about GPL regarding the stuff I do. I generally give the customer all of the source code anyway. If he wants to give it to the rest of the world that is ok by me.. Most of them machines are one-off machines.
I generally just get paid to make it work.
that's the best way :)
I think so.. I never made any significant money selling software anyway ...
By that I mean buying software, marking it up and then selling it again with the application
For most CNC Mill/Lathe apps, it seems like Axis works quite well.
yep. that's what it was designed for
anyway. bbl. have fun
mshaver .... was Axis around when Smithy decided to create Eztrol?
I just sent Smithy an email inquiry asking how I can obtain a copy of the source code for their Eztrol software - I said that I heard it was GPL software. No names were mentioned in my inquiry, and none will be mentioned if the discussion with them continues.
I'll let you know if and how they reply.
Dave911: Maybe, I wasn't around at the beginning of eztrol. AXIS wasn't always distributed with EMC, it started as a separate program that you could download from Jeff's blog/website.
that's true but it was always Free.
True. My point was really that the originators of eztrol might not have known of the existence of AXIS at the time. I didn't see eztrol until I started working for Smithy several years ago, and by then it had been in development for some time.
mshaver: ok ... thanks .. was Eztrol ever part of the download source for EMC2, or in the development tree, or has it always been a separate development? SWPadnos mentioned that he thought the source was on the server some place at one time.
Was Ajax CNC launched from the NIST code also? Or was that a totally separate development?
Dave911: At one time I tried to get eztrol to be part of EMC, but the board (perhaps wisely in retrospect) decided against it. I did send the very old code I had for them to review, but I never made it available for download anywhere.
Dave911: Centroid had a DOS based system before they switched to Linux. I would guess that they ported their original code to Linux. I think their system is a "smart box" type system, so I don't know how much value they would have gotten from EMC code. If they are at the CNC Workshop this would be an interesting question to ask.
>> but the board (perhaps wisely in retrospect) decided against it.
Why do you say that?
Centroid... Oh... I thought they were doing loop processing in Linux also.. I didn't know they used to be DOS.
Dave_911: Several reasons: 1. It requires QT4 which is not installed by default. 2. The developer is not an active EMC developer and they were concerned that if the project was orphaned there would be no one to maintain it. 3. The copyright status was uncertain.
Wow - must sleep! Good night.
Interesting ..... sleep ... yes.. I'm trying to quit.. wastes a lot of time ;-) Good night
Lerman_ is now known as Lerman
Lerman_ is now known as Lerman
There are a lot of bugfixes mounting up in the 2.4 branch. Anyone have thoughts about whether I should shoot for this weekend or next weekend for a 2.4.1 release?
imo there should be 2.4.1 this weekend, and after that if there still be many new bugfixes, 2.4.2 in next two weeks
has anyone built and installed a new packaged and really checked tooledit and tkemc?
micges_work: will you be able to let me know about the hostmot2 stepper changes today or tomorrow?
jepler, I'd like to "tag along" on a release, do you think you'll have time to explain what you're doing this weekend?
I'll be out of town next weekend
I'll organise some machine for test for tommorow
SWPadnos: I can't predict exactly when I'd be working on a release, beyond saying saturday is more likely than sunday.
hmmm. ok. Saturday is less likely to be good for me than Sunday
jepler: mail me if you'll find any more test
there are a couple of guides you folks have written that I should re-read
hmmm. I wonder if I have (a) a hardy machine or (b) a machine with VM software on it
that works reliably
cradek, that's incremental from the specified point, right? (as opposed to the current point)
no, I think it's from current, like an arc
SWPadnos: I tried at one point to make restarts reliable but I don't remember if that involved a different script name
I noticed a few scripts in there, and there are functions in the new-logger script (running, maybe_run)
but new-logger.sh isn't executable, so I wasn't sure which one is the "right" one to use
back to old logger.sh, with .jared removed
cradek, ok. I thought it was more or less a vector specification - a point and a direction/magnitude, so it seemed like it would make more sense as a point + offset from that point
SWPadnos: you'd have to weigh that against not making it otherworldly in gcodeese
jepler: running v2.4_branch I can't click/select a line in the preview
cradek: argh, I'll have a look
jepler: how would you feel about replacing the splash gcode with a pretty splined version?
or at least an arced version? it's currently pretty ugly
cradek: click to select works for the 2.4 I have here, but that's 88 commits before the present
cradek: I'd meant to replace with an arc version but never got around to it
I'm not confident enough in splines yet :-/
maybe saying something like "EMC/AXIS v2.4" would be fun too
and use a star trek looking font .. emc 2.4 is from the future!
oooh it's christmas
i've done some testing with three builds -- install,rip, simulator with tkemc,axis only
dgarr: reading just the commit message so far, as long as it's not because of hardcoded paths like "/usr/lib" it's OK if different values of --prefix don't all work. As for naming, it seems like tcl packages are usually named with initial caps .. is that what you saw too?
yes -- i used iinitial cap: Emc,Hal
using a different prefix only requires appending auto_path (i think) so seems reasonable -- i used it for testing to avoid clobbering /usr
dgarr: it was necessary to add these?
+ package require Tk ;# needed for font command
is this actually still necessary, or isn't it done by [package require Emc]?
jepler: what's your working ref?
cradek: sorry, I got distracted from that
cradek: looks like I tested at 3d01e8e
package require Tk was needed for an external tclsh program that used emc::standard_font_stuff
but I also just tested at c33d8d61 and it's OK there too
huh, c33d does not work for me.
package require Emc doesnt mandate Tk (i dont think a tcl Emc package has to use Tk)
to use emc::standard_font_stuff, the application should package require Tk earlier (or a Tk window will pop up on first usageof emc::standard_font_stuff)
it seems reasonable to me that external tclsh programs would want package require Emc and never want Tk
dgarr: if the program hasn't required Tk on its own maybe it's better to error out than pop up a window...?
on the other hand, it doesn't seem like it'll hurt anything
jepler: still no wfm
i like the error-out suggestion, i'll work on that
dgarr: with no change in that function you'll get an error (unknown command 'font') won't you?
yes-- error: invalid command name "font", if you think that is sufficient, i'll just remove the package require Tk lines
I'll go ahead and knock those out myself
jepler: they cyan highlight does work if I click in the gcode, but I can't click in the preview.
cradek: works for me. I tested lines (splash gcode), arcs (arcspiral), and splines (butterfly)
jepler: well wtf.
this machine is 10.04, sim, nividia, no desktop effects
dgarr: did you test debian packaging after your tcl package require changes?
no -- i dont know how to do that
you have to have the docbuilding stuff installed. then you can get the binary package of the moment by something like: debian/update-dch-from-git; debian/configure sim; dpkg-buildpackage -B
(or debian/configure -r to select the currently running realtime kernel, or -a, or specify kernel version)
cradek: some specific gcode?
it's cradek's "focus follows mouse" setting. metacity sends a motion event immediately after the button-press event. axis interprets this as a drag (to pan or rotate), so it doesn't perform selection
actually it sends two
the buttonpress and both motionnotifys have the same coordinates
so now, even turning off desktop effects doesn't give you a properly-working X server
EMC: 03seb 07master * r5d0e2493f43d 10/.gitignore: don't ignore the debian/extras/etc/modprobe.d directory
will you be on emc fest?
i'll be at the cnc workshop, for at least the last several days
are you coming to that?
unfortunately no, I'm afraid of fying
asking becouse of todo list in hm2 driver
what are you planning to improve (add)?
hmmm. which days do you expect to be there, Seb?
i'll probably be there wednesday 23rd through saturday 26th
i'm not sure what i'll hack on there yet
if there are any bugs in existing hm2 code i'll hack on that
i'm not in a big rush to do the hm2 todo items at this point, other than maybe spi
i might hack on heeks instead of emc2
* seb_kuzminsky ducks
bummer about the dates. I need to leave on the 23rd or 24th
23rd if I want to say hi to my wife on our anniversary :)
you know, in person and all
send her a text message
I suppose I could call her, but still :)
EMC: 03seb 07master * r2476c90e230b 10/debian/.gitignore: ignore the install dirs for the doc packages
EMC: 03seb 07master * rfdd7a696eb40 10/debian/extras/ (emc2.files etc/modprobe.d/emc2 etc/modprobe.d/emc2.conf): rename the modprobe config file to work with lucid
I'm heading to LAX on the 25th, so a few extra hours at home might be nice
dear mr 2.4 manager, please consider applying these three patches to 2.4
or approve it and I'll do the cherrypicking
i guess 2476c90e230b does no good on 2.4 (since it doesnt have the docs split out to doc packages), but the other two fix a run-time warning from modprobe on lucid, without breaking it on hardy
micges: did you get a chance to test the hm2 stepgen fix i did recently?
I'll do it tommorow
hi cradek :-)
wish you could come earlier... swp, jmk, jeff and I will be screwing around the previous weekend
sounds like fun, i wish i could come too
what'll you guys be doing?
seb_kuzminsky: let me look at fdd7 real quick .. if it's as obvious as it looks, it should go in 2.4
not sure yet. definitely going to HGR and maybe also Fair Radio which is not too far
probably hacking and/or BSing at jmk's house
sounds fun :-)
seb_kuzminsky: frankly I'm baffled by the 2476c change. aren't the same items just two lines lower in that file?
oops did i mess something up?
jepler: tried to build debian package -- i think it worked except signing, updated patch for rules.in, notes here: http://www.panix.com/~dgarrett/stuff/b3.log
ah oops, my bad
i was doing some of this hacking on 2.4, and i got myself confused... :-(
dgarr: probably you need to include those files installed to usr/lib/tcltk, not remove them to keep that check from complaining
* seb_kuzminsky kicks CIA-2
dgarr: oh, and yeah -- update-dch-from-git requires git 1.7 or something. if you care, you can get a newer git for hardy from this ppa: https://launchpad.net/~git-core/+archive/ppa?field.series_filter=hardy
i dont know what you mean by "include those files"
list in debian/emc2.files.in
and the rm for them is not required then?
a bit of background will help. hold on while I type.
we split the packages installed by the "make install" rule into multiple parts. In 2.4 the split is between the main package and the "-dev" package for building components
the way this happens is by listing them in a debian/*.files file. According to the files and patterns listed there, dh_movefiles moves the files from debian/tmp to debian/packagename
any leftover files in debian/tmp indicate additional files added to "make install" that need to be handled by the debian packaging
if you simply rm a file, that will make it not be included in the package
EMC: 03seb 07master * r1d75213d59f4 10/debian/.gitignore: Revert "ignore the install dirs for the doc packages"
if you did dpkg-deb -c emc_?????.deb | less and looked for the files you wanted in usr/lib/tcltk, you won't find them
I'll take care of the packaging later when I'm at a machine where I can install and test.
meanwhile you can re-make the tooledit patch using "package require Emc", maybe with a "catch" in case you want tooledit to work on a machine that's otherwise free of emc..
thanks -- i'll stop flailing at packages, I'll resubmit the tooledit patch after git master is updated
dh_movefiles: Compatibility levels before 5 are deprecated.
find: `x': No such file or directory
if [ `find debian/tmp -type f | wc -c` != 0 ]; then \
huh, what is that message in the middle from?
jepler: I didn't find any docs, how can I build emc 2.4.0 without docs?
Are you asking where it's documented how to build emc 2.4.x?
that too, I can manage to build debs on rt, I want to build emc-sim without docs if this is possible
oh -- no, the package build always includes building the docs.
oh ok, thanks
cradek: looks good to me. if it works in click to focus or pointer focus both, go ahead and push to 2.4.
jepler: wonder what other surprises await us
EMC: 03cradek 07v2.4_branch * r61eba92618e7 10/src/emc/usr_intf/axis/scripts/axis.py: Fix selection in the preview on Lucid
Smithy responded to my request for the Eztrol source code... the women who responded copied my request to their programmer .... so we'll see what happens next. :-)
Any bets ....?
are you aware that if they did not distribute the binary software to you (e.g. you bought a machine) there is no requirement that they send it to you?
That's what I thought... but I had to ask anyway...
The worst it can be is a "no" or a "hell no" .... and I am used to getting that response :-)
cradek, interestingly, that's not always the case (that they don't have to give the software to anyone who asks)
as jepler pointed out yesterday, section 3b of the GPL v2 says, as one of their options, that they have to provide the software to any third party who asks for it
their other option is to provide the code or a notice
their other option is to provide the code along with the program
(option 3 probably doesn't apply, since it's only for noncommercial distribution)
err, option c
But which GPL license would they be using .. if they say they are using one ?
I believe that one version of the GPL license is as cradek says .. at least that is what I remembered reading once ..
emc2 is GPL v2, and I think the Qt license is substantially the same
not that I've read the QPL
ok, I like my 6-core machine :)
real18m34.250s for a complete compile and package build of the ubuntu kernel
fast CPUs are ... fast
mozmck-6core, did you get any special hard disk, just upgrade CPU/MB/MEM, what?
I put together a whole system. I got the biostar mb with the 890fx chipset, 4gig DDR3 1600mhz mem
hitachi 500gig sata2 hard disk.
ok, so nothing really special there
I think the hard disk is the slowest thing on the computer.
sounds like it :)
what would be special?
one sec, I'll show you
A far as I've heard the hard drives that claim 6 gig/sec can't actually physically do it.
[23:41:35] <SWPadnos> http://www.newegg.com/Product/Product.aspx?Item=N82E16820148349
but ssd drives can ...
there's also a 128GB version, which only roughly doubles the price of the machine :)
heh, I'm still hurting from the damage does by this machine :) That's way above my pay grade :)
that one is the fastest I've heard of, at 355MB/sec reads (one article I read said that the drive actually did closer to 365MB/s)
wow, that would probably speed up everything a lot.
disk reads and writes, yes :)
one of hte customer reviews says something like "boots Windows 7 in under 10 seconds!!!"
A lot of that happening in a compile for sure.
hmm, he should try ubuntu 10.04 :)
yeah, that's about 10 seconds for me on my laptop anyway
maybe 15 from cold start
I think it's faster on this machine, but I think plymouth doesn't like my video card or something so that slows it down.
Asus card, nvidia 8400gs chipset.
I wonder if plymouth gets confused about what driver to use - vesa vs nvidia-specific
As far as I can tell so far, it looks like leaving the preemption model the same (voluntary I think) instead of CONFIG_PREEMPT in the kernel actually gives a little less jitter on the one machine I've tested it on so far.
same as the generic ubuntu kernel that is...
I can see that happening, until you're under load
then voluntary may get worse
Well, I was loading it both times. Running glxgears, starting and stopping firefox and other programs.
the machine doesn't do real well anyhow. about 41000 jitter with preempt, and 38000 with voluntary.
I wonder how to actually load the CPU, and more specifically the stuff in the kernel that might behave differently in those two modes :)
huh. that seems pretty bad
I'm going to try a non-smp kernel on the machine tomorrow and see if it does any better.
non-SMP and voluntary should be pretty bad (I think ??)
yeah, I've done everything I can think of in the bios and elsewhere. I have to load rtai_smi or it really gets bad.
This is a single core pentium4 machine at work I'm talking about now...
oh. I was wondering :)
3.2 ghz I think.
:) haven't tried one of my kernels on this 6-core yet. Just got it up and compiling this morning before going to work.
bummer, there's a limit of 5 per customer of those SSD drives on newegg...
yeah, I was so hoping to make a SATA-6gb RAID of incredibly expensive SSDs