some folks can't troubleshoot their way out of a paper bag
is the idea of following a signal from its source to see where it goes wrong really that tough to comprehend?
* jmkasunich goes away (dinner with friends)
jmkasunich: next he will ask about his following errors
that's where we started around 2pm
that was an odd ini file
did he ever paste the actual ini file?
I thought it was this http://www.pastebin.ca/327187
oh right, I remember the scales now
20, 40, 9.657170449, and 200
I think he changed pretty much all the numbers
the rotary maxvel is 2.5 (degrees) per second
with limits of -2 and +2 degrees
I think we should ask for a picture of this machine :)
I guess we don't have a 4 axis mm config
hmm where but the internet would I find a motorcyclist doing wheelies at 200kph
... in traffic
hmm he hits 300 sometimes
while going between two lanes off opposing traffic
darwin award winner at some point
unfortunately he'll probably take out someone else too
an inocent dar-standard..
again - sorry about the spelling
usually I can guess what word you mean :-)
trying to somehow relate it to darwin - sad
wonder why I got that video searching for "5 axis"
did you see the 5 axis machine that was making a compressor wheel for a tubo?
pretty darn cool
thought that looked odd
I can't wrap my head around setting something up like that (4 or more continuous axises)
I can understand controlling it, but probably not programming gcode for it...
not by hand anyway
all short line segments anyways. I think
I really need to get into 3d modeling. I had downloaded a free one that anonimasu had recomended but have not had tome to play with it. (I really should get back to trim ;))
what package is that?
alibea design.. I don't know if it is linux also.
ah I remember him talking about that
[01:18:44] <skunkworks> http://www.alibre.com/xpress/software/alibre-design-xpress.asp
looks pretty cool - again I am an 2d autocad person. ran a laser for a few years.
Someone earlier talked about highlighting a HAL file? I've got a really crude one running here now.
it seems that either bash or ini hilighting work somewhat
The highlight files are located in /usr/share/gtksorceview-1.0/language-spec
ah. have you modified one, or found one that almost works?
I copied ini.lang to hal.lang.
Then modified the first couple of lines by replacing ini with hal.
That succeeds in adding the hal highlight to the gedit menu system.
great. the next thing is probably to add the keywords and special symbols like <= and =>
It looks to me like the language.dtd
lemme check on that machine
is the key to the name of the element that starts a specific highlight.
The hal.lang header lines even allow you to select which type of highlight group it will be placed in.
In the menu system of gedit.
right - numbers, keywords ...
There is the whole regex thing available.
I highlighted loadrt and addf in red. Left most of the other commands green.
and gray/blued out the comment lines.
yep. I see that it assigns everything at the start of a line as a "variable" (in the ini version)
The other .lang files look like they can be instructive.
so the comment char needs to be changed to # from ;
C or tcl would be good, since there are language elements that go on the same line
HAL should have the keywords (at the start of a line, possibly with some whitespace)
they should be explicitly listed, because "frodo" as a command will generate an error
(at least for now ;) )
It doesn't look like it would be all that difficult to use. Even for g-code stuff.
G-code has some problems that are a little harder to deal with
spaces getting removed is one of them ...
I remember that from the kedit days.
yeah. Paul had soemthing that almost worked at Fest a couple of years ago (which I think you had worked on as well or something)
It looks like you can use as many keyword-lists as you like and just name them different.
Right I believe it became a part of the BDI about 40 or so.
so the commands and types can be colored differently
yeah - I saw it at the last NIST fest (and my first :) )
so newsig could be one, linkps linksp another
well, I'd make all commands the same color
Oh. We had just finished it a few days before that meeting.
data types another color
Gotta figure out how to get to an ini reference and make them unique.
oh, and categories something else (thread, sig, param ...)
hmmm. I think the kate "settings" type is a little closer - it recognizes # as comments
So you would make setp different colored from other commands.
no - setp, sets, show, newsig ... all the same color
but a show command has several parts: show <category> <searchspec>
those 3 things should probably be 3 colors
the more important thing is to show errors: if you have "show" followed by anything other than pin, param, sig, funct, thread, all - then it's an error
halcmd show pin ppmc.0.encoder ...
How often is show used in a .hal file itself?
that was an example
how about newsig ...
if that's followed by anything other than a type (HAL_S8, HAL_BIT ...), it's an error
(I may have the order wrong - the name may be next ...)
I guess the separation into a set of specific patterns is the most important.
BTW I liked your review in DM.
the biggest benefit isn't to be able to way "find the green 'setp'", it's "is there anything in bright red? ...
lagging slightly behind yor offsetting article :)
np it's getting late for me also.
it was funny to look at the TOC and say "I know that guy, I know that guy, I know that guy, I am that guy" ... :)
we need a regex that will find .[0-9]. and search for the start and end of the word (whitespace)
looking for a number, or for the instance number within a pin/param ...
within a pin/param
only highlight the whole pin param name.
ah, so a regexp that finds any entity that is separated by whitespace and contains at least one of the pattern .[0-9].
in the string element you can supply a "end-at-line-end=TRUE or FALSE"
I don't immediately see how to separate by whitespace, but not include the whitepsace in the found expression
maybe that's the \b I see everywhere
Yea. I suppose some nutcase will come along and edit the background colors and see our whitespace inclusions.
I know I tried it a bit ago.
I'd almost like to hilight the numbers differently so they stand out a bit
like make them bold or something
You are right the tcl.lang is interesting to see and see the effects of on a file.
man - the deafult colors in gedit are very distressing to my eyes
the fuschia and purple thing has got to go
That is one nice thing about the ability to alter them for your system.
heh - yep
But it also means that we can not write "look for the red words."
funny - the kate tcl spec doesn't corretly make "wish" a comment, whereas the gedit one does
err- the exec line that runs wish anyway
true, though those who do lots of customizations like that can probably be led in other ways
"red, or whatever you've changed the 'error' color to"
There you go.
Well that was fun. Thanks for the help, Steven. have a good rest of the evening.
its snowing :-(
it's just kinda cold here, but we have snow on the ground
couple inches here is all.
we got about 3 today
its falling pretty fast right now tho - the 3" per hour kind
hmmm. lake effect or a real storm?
fluffy stuff, won't amount to much unless it keeps going all night
not blizzardy, mostly falling straight down
ok. so we may not get that in a couple of days :)
I really like the ubuntu weather applet
I should do more manpages
but I want to about this ---><---- much
heh -even less, I bet
and I should be testing
and I should be adding to the stepper quickstart, and making a servo version
actually its quite satisfying when I finish one
actaully, I should be testing the iris on this lens
* jmkasunich nominates SWPLinux to write the manpage for modmath ;-)
I thought jepler volunteered for that one ;)
(he's asleep right, we can dump it on him?)
the ayes have it
what really bummed me out was when I realized that although I'm getting close to the end of the components list, the drivers list is untouched
the only drivers with manpages are serport and pluto
are they documented reasonably well in the manual(s)?
I kinda doubt it (some maybe, but some probably not at all)
some combinations of parameters of g76 will cut forever
SWPLinux: I just put all I know about highlighting in a wiki page.
heh - ok. I'll take a look once I've gotten this danged iris to move
noone ever cuts inside threads on a lathe, right?
Sure lots of folk do.
err - well ...
just don't try to debur it with emery wrapped around your finger.
Yep. After that we called him nine-finger-joe.
I take pride in that I have all of my fingers still. (knock on wood)
hmm, now I have mental pictures I don't want
see you guys.
cradek: does it miss the stopping point? (g76)
skunkworks: the whole thing is funny with negative numbers, it needs some more thought
did a == instead of a <= or something
for some reason that doesn't seem to be happening tonight - I think I'll go to bed
more like the loop adds when it should subtract or something
maybe someone will fix it for me before tomorrow
don't hold your breath
yeah I know
I'm too busy not writing man pages
we all have it a little rough don't we
good thing we get paid big
nah I got confused there for a minute
* SWPLinux goes back to the stupid iris
SWPLinux: need more voltage on that iris actuator
jmkasunich: I hope not - it opens just fine
if I want it all the way open ...
ah, it has become binary?
yes. no power = off, power = on ;)
you mean open/closed?
do you know what the actuator actually is?
it's a servo of some sort (like in model planes, not a motor)
ie, reversable motor, or solenoid working against spring, or voice coil, or...
no feedback other than the amount of light going through the lens
I think - the non-remote models of this lens have a "servo coil" and a "damping coil"
and no specs on how to drive them
* tfmacz says hi lawrence
* LawrenceG Hi Ted
XML is evil
and the tools that create it even more :(
the tool I
the tool I'm working with decides to write components semi-random into the xml file.. which makes diff & co useless :(
and it reindents the whole file.. which also makes cvs diff useless
I thought that was XML's advantage. Radom (randumb) placement of elements.
rayh: try checking for diff's..
I added about .5% to a 400k file
the diff is the whole fscking file.. I have no way of checking what i actually added
No kidding. I understand that you really need the parser and graphical interface to use it.
there is no way I can do this manually
It really is not intended for things like diff and cvs.
What are you working on?
it's supposed to be human readable
rayh: defining languages for AI software
[15:15:38] <alex_joni> http://en.wikipedia.org/wiki/Ontology
so.. are we ready for 2.1.0 ?
HAL:4: ERROR: module 'puma560kins' not loaded
any idea where I should look for more info?
nothing relevant there
is this sim or RT?
the module shows up under lsmod :(
Module Size Used by
puma560kins 21380 0
and nothing relevant is shown in the terminal?
(other than the HAL error)
emc_debug.txt and emc_print.txt are also of no help
the only thing I saw was that I tainted my kernel
yeah - I would have suggested those, but I've never seen useful information in them
SWPadnos: dmesg ends up in those
does the pumakins module return anything other than 0 on sucecss?
what's on line 4 of the halfile?
(I'm not sure how load errors are propagated back to halcmd)
does it call hal_ready or whatever the new scheme is?
juve@taurus:~/emc2.TRUNK$ bin/halcmd loadrt puma560kins
HAL:0: ERROR: module 'puma560kins' not loaded
I suggest looking at the code :-)
juve@taurus:~/emc2.TRUNK$ bin/emc_module_helper insert /home/juve/emc2.TRUNK/rtlib/puma560kins.ko
that works with no error
hmm.. maybe loadrt expects a hal_ready?
12:15:52 < cradek> does it call hal_ready or whatever the new scheme is?
I wonder if it needs to call hal_ready()
and hal ready
check for hal_ready()
another 100 point question
wtf is KINEMATICS_SERIAL ?
I don't see that in my source tree
I never said it's in your source tree :)
this is from emc1
in that case, it's an old thing from emc1
although I don't remember ever seeing it before
you're somehow particulary usefull today (like a fountain of wisdom)
a fountain of something anyway
whoa.. it starts
wasn't that supposed to be a framework/example of kinematics where one joint depends on another, like a robot arm?
though there are others like that I guess
but I don't see the need for serial
even trivkins is serial
e.g. not parallel
no - trivkins has orthogonal joints
moving X has no effect on Z
joint 3 was "-inf"
when I homed it, the display changed to "nan"
I guess it's got a broken arm ;)
that's the Puma saying "my shoulder hurts"
my shoulder is dislocated
I think it's a robot-mill combination
the shoulder is now a knee :D
hmm.. I am getting ferror's when I try to bring the machine out of ESTOP
maybe it's going towards -inf too fast?
cradek: does selecting a line in emctop work for you?
not for me
I doubt it ever has
another question.. jogging speed
(let me check something first)
ok.. I have 6 joints
all with the same settings in the ini
jogging however is different (faster for the first 3, slower for the other 3)
Issuing EMC_AXIS_JOG -- (+124,+24, +43, +3,0.000100,)
Issuing EMC_AXIS_ABORT -- (+120,+16, +44, +3,)
Issuing EMC_AXIS_JOG -- (+124,+24, +45, +2,10.000000,)
Issuing EMC_AXIS_ABORT -- (+120,+16, +46, +2,)
are your rotary and linear units both 1?
err are they the same?
but these are joints :)
you'll have to make them the same, since 012 are assumed linear and 345 are assumed rotary
one day this will be fixed :)
"works for me"
yay.. found some numbers where I can switch to world :)
X(350) Y(70) Z(-400) A(180) B(30.4221) C(0)
when you switch to world, do ABC disappear?
this is world..
joint mode is 0,1,2,3,4,5
oh they still mean something in world, right
axis shows them as joints though
sorry.. found the switch :)
jepler: wasn't there a version of AXIS that tilted the cone when A was changed?
yes, but we quickly decided that was no good
actually it rotated around +x
that's the most common :)
cool but assumes a lot about the machine
alpha (around X), beta (around Y), gamma (around Z)
at least that's how it works for robots
that's actually in the ngc spec
well the direction it faces is in the spec -- it doesn't say anything about where zero is
how do you mean zero?
on robots it works around the current point
east, north, ...
tooltip doesn't move
what is the "radius" when you're at Z0
cradek: radius is always 0
that's rarely true with actual hardware
but it's a bit more complicated than that I'm afraid
SWPadnos: it is true for a thingie called TCP
TCP = tool center point
you define that for the control
jepler: did you see the conversation with twingy last night around 11pm? I think he's going to make it build on ubuntu for the next release (maybe today)
sure, but it's translating that to actual joint positions that's the tricky part ;)
heh.. jogging joints is nice with robot kins :D
too bad the robot isn't displayed :/
cradek: do you remember what was changed to tilt the tip?
is it something I can easily do?
alex_joni: probably a rotation matrix in opengl just before the cone is displayed?
something like that
awallin: too bad you're not interested in robot kins :)
alex_joni: you have some nice robot to control?
but usually I have a control too :D
cradek: no, I didn't see that.
[19:00:51] <alex_joni> http://dsplabs.cs.utt.ro/~juve/dropbox/puma_kins.png
alex_joni: send me one of your robots and I'd be interested in controlling it! I bet it could mill MDF of some other soft tooling board? A 300x1000mm XY and 200mm Z working envelope would be just what I need
alex_joni: those circles are you jogging different joints?
awallin: the ones I work with are 5000mm diameter
alex_joni: sounds good!
400kg for the control
alex_joni: you could look in axis-historical to find the cone tip code - but it's just a call to glRotatexxx like awallin says
you just have to make the RIGHT rotate call
yeah, 3 of those
anyways.. I decided doing this blindly is not really possible
wonder if I could hook up a puma simulator to AXIS
found one in Java I think.. so AXIS could control that through jpype I think :/
seems like you could draw some lines that represent the arm
jogging is borken for nontrivkins
cradek: but as you said "works for me"
I thought it worked...
it "works" .. but speeds are totally off
I was in joint mode, and it was jogging joints 0-2 with the linear speed
do you want to jog joints or axes?
you can jog either
awallin: both :)
cradek: it's the assumption 012 = linear, 345=angular
I noticed another minor thing.. cone sits still.. velocity is !=0 and changes while I zoom in/out using the scroll wheel of the mouse
even panning makes it do that :D
I don't see it in sim/axis
it's probably because of the kins
position gets updated from joints with some rounding error
you can try this patch if you like:http://emergent.unpy.net/files/sandbox/joint-jogging-speed.patch
it is supposed to use "jog speed" for all joints, and the first 3 axes; angular speed for axes 4..6
it's hard for me to test it
I think angular speed for all joints is more appropiate
in this case
joints are rotaries :)
jepler: thanks for the effort.. but I'm gonna stop doing anything more on this...
fjungclaus-away is now known as fjungclaus
Where did you all get the time from for whole-day communication here :-)
Multitasking?? Somebody teached me, only women can do this. So Chris is Christine? :) :)
If I rember my beeing as a student (10 years before) there was some time. But now with job, family/children, house/garden ... there's n none
I now have much more free time than I ever had as a student
Ok, maybe I'm doing something wrong :) But I fear I have too much hobbies (cnc, astronomy/astrophotography + maintainig my own small observatory, auxilliary fire brigade, amateur radio, bicycling ...)
Maybe I should throw off some off them ...
hmm.. fireworks would fit nicely in there
you can watch them using your observatory, if something goes wrong.. you're on the fire brigade :D
Ok. I'll see if could bring the axis-mdi-history-patch to an end over the next hours ... so I switch over to something like half-away ...
No, firework Is very restricted here by the laws in Germany ... but we have an international approved fireworker here in our small village :-)) Sometimes he has a special permit to test some new rockets and the local fire brigade is doing the fire-watch then ...
Guest469 is now known as skunkworks