<jepler> "now how does this fit in (if at all) with the existing plots in axis?"
wanna talk about that?
sure, I'll probably be around for at least a half hour or so
I think it would be very neat to put the machine model in the AXIS preview window (optionally)
I'd like to retain the abilitiy to run a vismach window separately from axis (and from EMC)
but I also think it would be great to be able to replace the cone with a machine model
as for how to do that, I haven't a clue
I think I can handle that, technically
btw, I think I'd like to rename pumagui and scaragui
I think I can offer to keep the work, the world, or the tool tip "in place" as everything moves around
sure, I am not thrilled with the names
pumavis.py, scaravis.py, etc crossed my mind
also not sure where they should live
the source files? yeah, good question
I can easily see people hacking on them to customize for their machine
but we have a lot of freedom to move them until we release 2.2.
we don't want to do it too many times tho, cvs doesn't like that
(loses history I think)
so we should stew for awhile and then pick the correct answer
thats why I brought it up - to start the stewing
I actually thought about sticking them in the configs tree
put scaravis with the scara config
does that work out, in terms of halcmd's search procedure for 'loadusr'?
no clue ;-)
just tossing out ideas at the moment
when an ini file specifies a var or nml file, its assumed to be in the config dir
hal files two for that matter
but I don't know how loadusr works....
I think the current dir during startup _is_ the config dir
unless there's a bug in the kinematics module, the AXIS backplot and the vismach backplot would be the same, wouldn't they? if that's true, then for the usual case of machine-visualization-in-axis I'd stick to the current way of generating the backplot
so loadusr foo should search there first, then search some path
I don't know about that
(them being the same)
did you try tilting the scara worktable half way thru milling the splash screen?
you get bent words
how do I do that?
(which is what would happen on a real machine)
setp scaragui.0.joint4 <something>
I used touchoff to get the words about 1" above and approximately centered on the table
then tilted it while machining
vismach can properly display things like milling on the outside of a pipe, if the pipe is mounted to a rotary table and turns during cutting
no clue how axis would deal with that
(or emc itself for that metter)
I am now convinced the vismach backplot will be better for those cases
vismach has the advantage of telling you what would actually happen, but it really doesn't know how or why
a preview in vismach would be a nightmare I think
AXIS records position from the stat buffer in a thread that isn't even running python code
it would have to switch to recording HAL joint positions
I wonder what coordinate system that is in?
yeah, but what xyz?
machine base (the good old earth)? worktable? G5x offset?
(the g-code coordinates with g5x and g92 offsets applied)
iow, "this is where you told the tool to go (or more precisely, this is where the TP told it to go, with corner rounding, etc)
any errors or such in kins would not show up there
vismach says "heres where it actually is, after kins"
ideally of course they match
but for 4 or more axis work, the vismach approach is much simpler
at the expense of not being able to do previews, or limit checks, or check the extents of the program, or.... a lot of things that axis can do
this might well be a case of two different tools for two different jobs
if you have real iron, vismach has little or no purpose
I hope it doesn't sound too cynical to say that I almost think of this as a marketing feature
if it lets us get a better grasp on dealing with >3 axis machining, that will be good enough for me
basically, with only a little work, every developer can have a shop full of machines now
rotary axis mounted on a bridgeport table, sure, we can do that
gantry mill with a 2 axis gimbaled cutter, we can do that
I'm quite impressed with the "look" of the machines you did so far
pov-ray experience kicked in
basically, vismach gives us a limited but capable subset of pov-ray's language
and because it uses python, you can do a lot of custom stuff as needed
yeah -- I always missed in POV the ability to drop into a "real" language as needed
they tacked on some kind of macro thing in the later versions after I quit trying to be a raytrace artist
the language has really grown - its a complete programming language now
but thats an aside
for our needs, this approach is better - we added the shape and transform primitives to a real language
btw, thanks for the time spent walking me through stuff last night, it really made a difference
I guess I need to figure out how to make vismach able to take "live" HAL data for the standalone, or "canned" data for integration with AXIS
can I nag you about something?
sure, what's that?
you offered to make scaragui get D1, D2, etc from the command line
I don't know how to do argv/argc stuff in py
and although it would be a good learning expernience, I need to sleep one of these nights
sys.argv is like C main's argv .. len(sys.argv) is like C main's argc
but my offer to do it still stands
do you think that anyone who customizes it would change all the numbers, except in rare instances?
I've really enjoyed working on this - learning a new language and doing the 3d matrix stuff that I haven't toucned in about 10 years has got the brain juices flowing
so I don't need to provide a way to customize just d4, for instance?
but I gotta stop staying up so late
I was reading about py function calls
there is a syntax where you can do func(foo=3,bar=5) and all the others use their defaults
does that seem like a good choice for the command line? or should is just be "scaravis 123 543 423 125"
I was going to write "scaravis 123 543 423 125"
but scaravis d1=123 ... would be more documenting, assuming that 'd1' means something to a person with a scara
and it should
[03:04:26] <jmkasunich> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/emc/kinematics/scarakins.c?rev=1.4
they are documented there
at some point as the scara config stabilized, that should be pulled into lyx
maybe we could even take a screenshot of the simualtor and use gimp to draw the dimensions right on the machine
my straight lines aren't straight! (when the simulation's dimensions don't match the kinematics)
I think if you scaled both arm lengths by the same amount, you'd get a scaled but straight result
but if the ratio of their lengths changes, all bets are off
loadusr -W scaragui d2=200 d4=300
OK I'll do it like this
special bonus: you can write 'd2=3*4*5' if you want
they are treated as expressions?
for setting in sys.argv[1:]: exec setting
does that mean you could even do "scaragui foo=30 d2=foo*0.5 d4=foo*0.8"
there's a way to say "if this_var doesn't exist: set this_var", isn't there?
I side-stepped that by putting the defaults above the 'for ..: exec'
oh, I was thinking farther than that
but yes, there is a way to find out if a global or local variable exists, similar to what I did with hasattr() in vismach
[03:18:30] <jmkasunich> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/emc/usr_intf/axis/scripts/scaragui.py?rev=1.5
see the block that starts with "# calculate a bunch of other dimensions"?
I was thinking of making those conditional (if the value exists, don't calc it)
then they could use your exec trick to specify additional params, not just the 6 critical ones
want fatter arms, just add l1_dia=whatever to the command line
>>> 'test' in globals()
>>> test = 3
>>> 'test' in globals()
is not ! ?
no, it's 'not'
if not l1_dia in globals:
l1_dia = d2 / 5.0
if not 'l1_dia' in globals(): l1_dia = ...
oops, forgot the quotes
tool_angle = math.atan2(d6,d5)*(180.0/3.1415927)
by the way, the constant is math.pi
I figured there was one, but a quick search didn't find it, and it was late
nobody to ask
and actually there are convenience functions math.degrees() and math.radians()
feel free to clean up the cruft
actually I think I'll feel free to call it a night
feel free to clean up the cruft some other time ;-)
see you around
actually, at some point I'd like to explore making classes for 3d manipulations
I have to imagine some have already been written
my invert function should be a method of a matrix class
I did a lot of googling for them last night, but I was looking for the code
didn't know how to import a lib even if I found one
anyway, thats a topic for another time
>>> from numarray.matrix import Matrix
>>> a = Matrix("1 0 0; 0 2 0; 0 0 1")
(requires package python2.4-numarray-ext)
cradek_ is now known as cradek
rayh: seen the robot screenshots?
* alex_joni started to enjoy machining with emc2 :D
rayh: scara: http://axis.unpy.net/01170693566
alex_joni: it's about time!
"and you can sit and watch as the hypnotizing robot arm mills your nc file in its virtual world." <- nice :D
I just wish it moved faster
I ran the scara robot a bit yesterday.
alex_joni: I wish the scara moved faster
Wanted to do some testing with g55+ and C offsets.
I suppose I should try changing the ini file
jepler: that's teh trick :D
btw.. I forgot to commit some sane TRAJ/ ACCEL for scara
will do that now
with that low accel, the robot moves really sluggish..
jepler: should be better now
I forgot -- scara doesn't even run here, because I put some 'limit' blocks between commanded and feedback position, and I get following errors
guess I should take 'em ut
hmm.. I don't have those
maybe I didn't check out the latest ver here
no, this isn't checked in, it's a local change
ahh.. ok :)
cp -rf ..
cvs co -dPC :D
cvs server: cannot find module `:D' - ignored
did you 'apt-get moo' today ?
What's an 'apt-get moo'?
* rayh wonders how much he trusts aj!
* alex_joni grins
rayh: you can do it without sudo
so even if you don't trust me.. it won't fsck too many things up :P
did the parport image go away in the current docs?
it's busted in the HTML version of the docs for some reason I don't understand
as far as I know it's right in the PDF version
I don't see anything but xx.hal as a literal in the user manual now.
"parport block diagram" is on page 54 of http://linuxcnc.org/docs/2.1/HAL_User_Manual.pdf
damn it's getting hard to find stuff that I've become acoustomed to attaching to help mail.
anyone with emc2.1 installed from the package will have that diagram on page 74 of the EMC2_User_Manual.pdf which is available from the gnome menu.
er, page 76
I guess that's not the drawing I was remembering from previous manuals.
This one is emc2 specific. It does seem to have a great deal more detail.
I was able to find what I wanted in cvs in the old documents module.
I don't believe anything's been removed completely, but it's possible it would be in the Attic/
IMO the users manual needs to be de-hal'd a bit.
It reads easily for those who know what something like (bit) parport.<portnum>.pin-<pinnum>-out means.
But a simple listing of the effect of standard_pinout.hal in a table form would be a lot easier for a new user to understand.
maybe you were looking for this image? http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/~checkout~/documents/images/pinout.png?rev=1.1;content-type=image%2Fpng
and there was a table listing standard pinouts right after it in the lyx that used it.
I got em. Thanks for the help and for letting me vent.
everyone agrees the docs should be better
you have a perspective on the needs that the rest of us may not
Ah. I see that the m5i20 has tables.
Perhaps what is missing from the parport are similar tables for standard pinout and perhaps xylotex pinout.
rayh: might be :)
Let me try adding those standard configs in tables and see how it looks.
you'll add them to the chapter called "basic configuration for stepper systems"?
config/stepper.lyx I think it's called
I see it.
jepler, What gtk2 do I need to compile ladder and such?
rayh: I've compiled it with versions as old as gtk+-2.4.17 but breezy and dapper both have 2.6
looks like ubuntu calls the package libgtk2.0-dev
The configure fails to find gtk2 on the latest cd install.
it should be installed by 'apt-get build-dep'
that command complained only about lyx2html
latex2html is the only complaint. Sorry
sounds like sources.list still points at the 2.0 repository, not the 2.1 one. Is this the new 2.1 install CD?
$ apt-cache showsrc emc2 | grep ^Version
do you get something different?
I also get the same set of errors regarding tcl that were mentioned in email.
Yes it's the new one.
No answer to the apt-cache.
does the included version of emc2 run?
or does it get that tcl error?
I'll try the new install/live cd in a vmware tonight and see if I can duplicate any of these problems
but the one you built yourself gets the tcl error involving "msgcat"?
The config command came back with no tcl.h ...
I've not tried to compile yet.
I'll try an apt-get update and recheck the repositories.
oh -- you're not talking about the problem reported by Larry D. O'Cull on the mailing list yesterday, you're talking about configure failing
though I guess he got to the same problem later on
I wonder why apt-get build-dep is broken on the live cd
pastebin your sources.list please
and the output of 'apt-cache showsrc emc2' if you would
[17:15:10] <rayh-cd> http://pastebin.ca/344253
I did remove the source repositories except emc
and 'apt-cache showsrc emc2'?
nothing jumps out at me as obviously wrong
in the sources.list
you never completed 'apt-get build-dep emc2' because of this error involving latex2html
it's no wonder the following steps fail
okay. I don't find latex2 in dapper
add to sources.list: deb http://us.archive.ubuntu.com/ubuntu/
sudo apt-get update; sudo apt-get build-dep emc2
[17:21:50] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=32001
skunkworks, I'm pretty sure that developer would be Paul, and I'm not sure what the end result was
I think he was working on the Vigilant drivers (or had just gotten them to work) about 2 years ago
there's a driver for the "PCI-ENCDAC 4" (hal_vti) but all I can tell you is that it seems to compile
yeah - I don't know what features of the card(s) are supported (or not)
72.3 meg of additional download!
Never gonna happen.
heh - see you in a few hours(days) :)
do you have a DVD-rom in that machine?
I don't believe that I've had this kind of an issue before with building emc2.
are there additional dependencies or what?
how would DVD-rom help me now?
it won't help right now, but I'm curious as to whether a DVD with all the build tools would be a generally useful thing
Could someone write up a reply? Or ignore? :)
ah. This box does not have a DVD.
if you have a machine that can compile emc2 2.0.x, you can copy the .debs (from somewhere in /var, I think) to a CD/USB stick, and manually install them
I would get one if we had such a disk.
unless it's new stuff that wasn't required before
Although a two set cd would do me as well.
SWPadnos: there are new requirements to build emc2.1 compared to 2.0
but I don't think the live CD had the build tools either
No but it was not this much a problem when I installed from the old cd and built emc2 about a month ago.
do one of you board members want to contact this Vigilant Products guy? otherwise, I can do it
I think their cards are a "true analog" version of the Mesa - DAC output and encoder input (plus some extra I/O)
I don't care to be in a position of apologizing for um .. someone else .. being hard to work with
or not keeping up his end of a deal
that's one expensive card
It doesn't have to read like an apology. "That developer left the project over a year ago."
$1000 for 4-channel DAC+encoder + 8 aux I/O
One of the Vigilant guys attended the first fest at Detroit.
Jim something if I remember at all.
this e-mail is from a "John Rowland"
[17:39:07] <jepler> http://www.cnczone.com/forums/showthread.php?t=32001
if you missed the url when skunkworks pasted it
He sent a card set to Paul for writing the drivers for emc.
I could have sworn those were PC/104
maybe that was a generic digital I/O card Paul had
SWPadnos: the post says "boards" (plural)
jepler: you gonna answer that?
maybe suggest that he direct questions to the emc-developers and/or emc-users list ...
* rayh twiddles thumbs waiting for cnczone amid the massive download.
or emc-board, it shouldn't matter..
I wonder if a private contact first would be best
alex_joni: I was hoping someone who could sign as a member of the board would contact him
but I wonder is it that hard to contact us?
jepler: I think I can do that by mail..
I read that..
TURN THAT BULLSHIT OFF
eek.. I had no idea that was there..
oh good effing grief
I guess people would complain if they didn't do that too.
So.. I don't have to worry about it? :)
thanks for spotting it
jepler: I would gladly turn that off, if I only would find how..
maybe your google skills are better than mine :(
jepler: can you check if it's better now?
looks good to me
there's a space at the beginning of the address but it sholdn't matter
alex_joni: you are going to contact the vigilant guy, then?
I think I should. I remember the relationship a bit.
the board assigned ray for the task
but I would like someone with access to that forum to reply there too
say that "thanks for letting us know, the board was notified, and will reply by email soon. the problems with cloaking emails on the website have been resolved now. ..."
to say what -- give the contact address, say the website's corrected, and that a member of the board will contact him personally in e-mail?
something along those lines
I can do that, I just created (or tried to create) an account on there
it doesn't let me post
I think greylisting lost the confirmation e-mail
skunkworks: so if you can post a followup, please do
is this ok then?
[18:30:37] <skunkworks> http://www.cnczone.com/forums/showthread.php?p=254293#post254293
skunkworks: yeah, pretty ok
SWPadnos: I remember the pc104 board too, but that was a Sensoray 526?
afaik, it's not by Vigilant
<- this one
Yep. I believe that Paul got both the vigilant and sensoray cards during trips to NAMES.
rayh: did you hear back from vigilant? (too soon I suppose) :)
Still composing the mail.
just sent the reply to vigilant.
seems we got the best man for the job
hallo Frank :)
wie geht's ?
Kam gerade von der Arbeit heim, esse zu Abendbrot und kann mich jetzt noch ein paar Stunden um meine Hobbys kümmern. Also gehts mir gut :-)
hey hey hey - stop that. be nice to the sub-mono-langual people
Alex started that ;-)
alex_joni, yep - that's probably the one I was remembering
is it just me or my isp or is email in a bit of a backlog?
email sometimes seems to act like an inchworm
bunch up, stretch out, bunch up, stretch out ... :)
I couldn't even get a working link to smtp for a bit there.
that's an issue with your ISP then, most likely
suppose we need to apply the notion of rush hour.
in da up 'eh.
heh - yeah. with lots of rubbernecking along the way
slowing down so you can look at why other cars are slowed down (to look at stuff such as accidents and things)
I got some modified kins from the guy with the bot in germany
maybe someone (with some math understanding) wants to look at them
[22:13:08] <alex_joni> http://pastebin.ca/344653
yeah, that's the common way to describe a robot like that
a scara is RRTR I think
T = translate?
"Find solution for control of a 4 DOF SCARA robot KIWI with RRTR configuration via Internet. "
seems I was partly right
what are the meanings of the alpha / a / d numbers in the table in the comments?
hang on, I'll make a new one (translated)
heh - thanks :)
SWPadnos: what's the base of the robot called?
below the first joint?
I don't know. platform or base maybe
base would make sense to me
get a clear GPL notice, convert the parameters into HAL parameters, and go for it
comments translated into english are a plus
[22:20:11] <alex_joni> http://pastebin.ca/344664
jepler: it's mostly the puma code
The design of that bot looked a lot like a puma.
yeah, but it seems the puma kins were designed to turn some way his bot can't or something like that
if you're concerned the new code is "too puma-like" then maybe we should make a single module cover both machines
jepler: right.. but right now I don't know which is the more general one :)
I'll need to read up on this topic..
I don't see a problem with different versions.
rayh: if they're 95% the same.. it's just duplicated code
In emc we had several. A couple were only for one-off machines.
* alex_joni runs to bed ..
good night all
Right. And if we had a way to adjust values after compile it would be nice.
see you Alex
rayh: they can be adjusted by making them HAL parameters
I can see params for limits and such but can it handle things like the nature of joints?
lengths of arms?
I think it's possible to do all that stuff with HAL params. I don't think it's the best solution though
there's no reason to have things like arm lengths dynamicallyy changing (unless that's a joint, which is another thing altogether)
In the existing kins like genhexkins the strut length is compiled in.
That would make the kins for each machine unique.
for the scara vismach thing, we (jepler) made the arm lengths and other params be command line args
right now it's easy to make a small number of items configured as HAL parameters, but it's true that there's no reason to change them dynamically
we could make them insmod args for the kins modules
there is not a float-type insmod arg
you wanna write 'atof' and get it right? not me.
I don't particularly like using hal params tho
I've gotten into the mindset that params can/should be dynamicallly adjustable
but really, things like this are what they were originally designed for
it might make more sense to swap a segment of the robot arm for another one without shutting down HAL, than changing the position-scale of stepgen (when it is running a mill) would
theres really nothing wrong with using params