03alex_joni * 10emc2/src/emc/iotask/ioControl.cc: added a function to set _safe_ hal pin values, on estop, on init, etc.
hello.. anyone still around?
* cradek cheers at alex_joni
cradek: wot's wrong?
thanks for cleaning up that crap
ahh.. you probably seen the $log stuff :)
cheers, not jeers
no idea what jeers means :/
jeer (verb) [I or T]: to laugh or shout insults at someone to show you have no respect for them:
The people at the back of the hall jeered (at) the speaker.
ok.. thanks for todays lesson :P
np.. we talked about this a while ago.. just had it in the back of my head
yeah, it's good to get rid of it before the plague infects more files
changed CodingStyle in the process.. maybe you take a look if it's ok?
strange, it's in src, not docs
well.. it's about src/
% wc -l CodingStyle
a bit long?
can we get that down to 66 lines or less?
if you can.. I have no problem with that...
but not wider than 255 chars :D
no, 80x66 definitely
heh.. ok, but there's a lot of info in there
yeah, I'm not entirely serious
but it looks a bit daunting
well.. I did recall I forgot how it started by the time I finished it the first time :)
yeah, and I find myself scanning instead of reading it
Implicit tests for zero should not be used except for boolean variables.
e.g. if (spindle_speed != 0) NOT if (spindle_speed)
um, because ... don't we all know C here?
The form while(true) should be used for infinite loops.
what's a true? There are several good ways to write that without depending on a define
Only loop control statements must be included in a for() construct.
(and the NOT example won't compile)
you disagree a bit much today :)
you asked :-)
I asked about the paragraph I put in there
about CVS $foo's
well, I must have scanned over it
CVS foo.. pretty much in the last 20%
CVS Special Fields
mind if I add a couple things?
not at all
je suis enchante
hope so ;)
did you test the stepgen one?
sure I did
works ok for me now
not so sure about the iocontrol stuff
* jmkasunich didn't know the syntax for array MODULE_PARMs
wanted to talk about you about that
* alex_joni neither
but the first thing I tried, seemed to work :D
I must admit I think Yabo's approach is nicer than the cfg=<some_string> method
about iocontrol.. what's your feeling about estop and hal-pins?
yes, easier to do/parse
maybe we should change them all like this
except the cfg="OOII" ones :)
and just as user friendly, if not more so
gets rid of those nasty _, , parsings
I've had problems with the quoted strings, when used in scripts, spaces within the strings sometimes mess things up
yes, I know
like the init_pins() fix
hmm, that wasn't very clear
I like the fix
the init_hal_pins() one
safe values are a good thing
yes.. not fully sure it doesn't block smthg.. need to simulate a machine with tool-changing again
but it's better to lock io up, then let it move freely as it was before
not sure what it was doing before
hmm.. mostof the values got reset
because of the NML, and the logic inside task
03cradek * 10emc2/src/CodingStyle:
Quick first pass fixing grammar and spelling. There is still some work
to be done. Also, I added another paragraph talking about cvs logs. I
think it's really important. Lots of programmers completely misunderstand
the purpose of cvs logs (in my opinion).
* jmkasunich wonders how hard it would be to make a little user space program that had two HAL pins... when pin A goes true, the prog pops up a dialog with a predetermined message, like "Please change tool", and when the user clicks "OK", it asserts the other pin
use that with your tool-prep and tool-prepped (or tool-change and tool-changed) pins for manual toolchange
actually it wants one more pin, with the tool number
not very hard
a simple bin/halcmd script might even do the trick :D
with message in it
[00:42:42] <robydeb> http://digilander.libero.it/informatico2003
robydeb: what about it?
alex: how would a simple script pop up a dialog?
well.. an xterm
jmkasunih I have begun now
alex, cradek - take a look at CodingStyle ;-)
it will become in future an area wi-fi
including $log$ in the file made a mess
because CVS interpreted it
ohhh.. bummer that
that of tasks of the Italian translation - English?
cradek: lol, that's why I added only $log to my paragraph, not $log$
I guess cradek gets to fix it then
* alex_joni and jmkasunich point to cradek
I have no idea what he was talking about ;-/
the URL was entirely in italian
didn't see EMC or CNC or anything like that anywhere
there was some english translation..
plenty of flash and stuff
but nothing worthy
some link to skype :)
alex_joni: oh nooo
alex_joni: I can't believe I didn't think of that
see what you've done? :))
ROFL, it can happen
to anyone.. even to you :D
maybe I unscrewed it
03cradek * 10emc2/src/CodingStyle: fix my screwup of the keywords
that was pretty easy - I checked out with -kk
Jacky^ is now known as Jacky^afk
* alex_joni goes to bed
anyone here used a Shopmaster Eldorado combo mill/lathe?
I have a Shoptask, a somewhat older version
worth a damn?
depends on what you compare it to ;-)
the lathe part isn't bad, the mill is limited
well, I've got a Bridgeport that I'm gonna update the CNC control on to EMC...right now it isn't usable (and doesn't even have manual controls)
and I was thinking about getting a mini-mill and mini-lathe and do the CNC conversions on them to learn about that part of the game
as a mill, it's not terribly rigid, and the lack of a knee means you're always blocking jobs up to reach the spindle
and then I'd have small mill and lathe around
and I've have the big boy for big jobs
if you have space for separates, that is probably the way to go
well, I do plan to do a good bit with plastics, too, so the small ones would be fine for that
is the eldorado the one that has a traveling head standard?
mine doesn't - and that sucks
if you can raise/lower the head, that eliminates one major complaint
looking at his blocking in the pictures of it cutting, I don't think so
[03:49:58] <djb_rh> http://www.throwarock.com/CNC4Sale/pages/IMG_8717.htm
that is very similar to mine, except for the tailstock
my tailstock is older, has very limited travel (barely 1.5")
maybe the mini-mill is the way to go
the Z travel on the Shoptask is only about 3"
I'm sure I can sell it after CNC conversion and not lose too much on it if I find I don't need it after I get the big boy working
heh - you can never have too many mills
he who dies with the most tools wins
anyway, my $0.02 about shoptask (having owned one for 7 years): If you have space for separates, get separates
if you need to fit a shop in a small space, the Shoptask isn't bad
BTW, I now own a Van Norman #12 mill and a SouthBend 13" lathe, in addition to the Shoptask ;-)
I still use the Shoptask though, because the two large machines are out in the unheated garage, the ST is in the basement where all my other tools are
ouch - he's asking $2700!
I paid $1500 for mine (no CNC though)
the Van Norman and South Bend put together cost less than the Shoptask ;-)
a new Shoptask is $2300 plus $399 in shipping without CNC
pricy for chinese iron
the CNC bits are gonna cost another $1200 or so, I think (if you figure $500 for a PC), maybe more
it's easy over $4k in stuff
and it's local
the one that guy is selling?
his is the "current" Shoptask model, from what I can tell
where are you anyway?
oh shit, it's at least $4500 in stuff
the ball screws for that thing are $799
I'm in central NC
the latest model has a head that can be raised and lowered, and a brace at the tailstock end that makes it more rigid
ok, NC isn't exactly used machine Mecca, is it?
crap, you're right
(I'm in Cleveland OH, where used American machines are fairly easy to find and cheap by comparison)
living where you are, and with the ST machine local, it doesn't seem like a bad deal
yeah, but I think I'd still rather do the mini-mill/lathe thing
I know they'll do less, but they're enough for much of what I need and will be good for learning on since they don't take much space, either
and they're cheaper
and I'll learn more
especially since you have the Bport anyway, you can get away with a little mill for now
for a lathe, don't go too tiny, cause you'll want bigger soon enough
I still think mini-lathe is fine since I'll just learn with it and then sell it after I get a bigger one
any "loss" I take in selling it will be cheap education, I think
assuming it doesn't remove any limbs. :)
if it's small enough that's unlikely ;-)
* jmkasunich knocks on wood
just digits, then?
well, you start with plenty of those
it's not like they are arms where you only have two
still wanna keep them all
yeah, I'm joking
hey, I race cars
I'll likely lose limbs that way long before any shop accidents
guy answered my questions on the shoptask, but I told him thanks but no thanks
he makes model ships
but found he couldn't make hardly anything he needed without a 5 axis mill
ah, that explains the wood dust on his mill
and he wasted too much time trying to draw the parts anyway
and why he doesn't worry about rigidity
all plastic and wood
could probalby do that with a sherline
can you get a 5 axis sherline?
I know they go four
I don't know, I doubt it
but I thought the fourth was a rotating table
don't think you can make the head turn
I doubt he really needs 5 axis, thats pretty radical stuff
hey, maybe he makes radical model ships
03paul_c * 10emc2-auto/wiki/ (25 files in 16 dirs): "Auto update wiki from a cron job. Mon Oct 31 05:30:01 GMT 2005 "
lerman_ is now known as lerman
* fenn whines
wot's wrong fenn?
why would making axes and joints work right take a lot of changes?
because of the datapath to send the information
you need new NML messages
but you dont need to send anything
yes you do
because you're in a special case
XZA not XYZA
and no-one reads that from the ini file, beside the GUI
where does motion number the axes?
in emcmot.c(emc1) or control.c(emc2)
but axes don't get numbered
they are always numbered the same
X=0, Y=1, Z=2, A=3, etc
so the problem is that joints dont get created 0 2 3
likewise for joints.. j0,j1,j2,j3 (in emc2)
that would be one way to handle it.. j0,j2,j3
the other one is the one I said, link x to j0, z to j1, and a to j2
either way is very simple to change in the source code, very hard to automate
pass the info from the ini file to the RT motion part
* alex_joni would do it like this:
1. copy trivkins.c to lathekins.c
2. modify it accordingly (like described in the email)
3. link lathekins inside the motion module
4. run it
would take .. maybe 10 minutes
hmmm joints are called axes inside of hal.. bleh
they are called axes inside the ini too..
although joints is more appropiate
why is it always such a big deal to add an NML message?
it's not such a big deal
but what's the benefit in this case?
complicating the heck out of the code, to suit another .1% of the users?
which could make the change easily themselves?
I'm not ok with that
what's not quite as obvious is that if you have a "c" axis it gets mapped to the a axis
how do you mean that?
well, standard axis naming convention goes "x is parallel to a, y to b, z to c"
that's how the interpreter assumes it
so if you have a c axis, then it will look like a is moving when it is actually c moving
depends on your kinematics file
and the set-up
if you have all 6 axes & joints, then "c" moving will look like "c" moving :P
but nobody has all 6 axes
ok.. then trivkins is not suited
in cases where some axes are missing
right, if nobody is ever happy, then it should be fixed
I am always happy
since nobody has a 6 axis mill (or do they?) trivkins is always busted if you have axes b or c
if you have XYZ then it's not busted
ok but you know what i mean right?
If you have special case machines (XZABC or XYABC or whatever) then it's not working right
but I don't see it as a bug
it's a misconfiguration
even if the configuration is inside a compileable file
that seems poopy to me
rant on the list :)
see if others agree
I agree it's poopy..but I don't agree it's smthg to waste effort on right now
read "other priorities on emc2"
right you gotta get hal/nml figured out
trajectory planner fix
tp fix / improvement
would it cause much of a stir if i went through and renamed all the wrong "axis" function/structs as "joint" functions
probably not eh
actually i bet john would throw a fit because it changes the ini file
hmm.. something fishy happened with my irc-sw
U h4v2 b33n h4X0r3d :P
[13:37] <alex_joni> there is also the concearn about fusion in the back of my head
[13:37] <alex_joni> seems rtai is dropping the current approach, and choosing a new way.. emc won't work on that as it is now..
[13:38] <alex_joni> lathe stuff
alex_joni_ is now known as alex_joni
I agree on the ini file...
it would cause too much confusion
is .ini file interoperability a goal for emc2?
(i realize this isn't worth breaking interoperability for)
dunno.. ask the board :)
hey you're on the board now right
NML interoperability is one thing I worry about
that's why I said ask the board, not me
maybe wait another week to ask the new board,
you'll probably not get an answer from this board
do you usually have to re-home a machine tool if you turn the power off? or mostly they have batteries or something these days?
depends on the machine tool
I can tell you how it is on robots if you like
the oldest ones were using encoders, switches and index pulses for homing
the newer ones use resolvers (they know exactly where they are inside a rotation)
like an absolute position encoder?
they used some batteries to hold position when the control was switched off
kinda, but only on the current rotation (other rotations were kept in software counters)
newer systems used a nicer approach:
on shutdown there was a powersupply that held the voltages for another 30 seconds (off some big caps)
and the current position was stored inside some eeproms
and read on the next start-up
the newest ones write it to some SRAM all the time
so power down shouldn't affect them
SRAM with battery afaik
but the best way would be abs-encoders if you ask me :)
well my motors already have incremental with index
then.. you could save the value on shutdown
given you have a normal shutdown procedure
but on powerloss you probably will have to rehome
also.. what happens if the axes/joints move during power-off
i think keeping the encoder chips on with batter power all the time is the best solution
maybe i will need some big batteries to keep 6 or 7 dsPIC's up all night
what good does it do if the encoders are powered?
they won't do anything..
just run the control logic off a separate power supply like you should be doing anyway
control and encoders
not really any reason to shut down the control logic.. you can run it on a ups if it's an issue
* fenn pats his dumpstered UPS
glad i sorted that out
how do you plan to build the machine?
octahedral hexapod with acme screws
i need to find a bunch of cheap angular contact bearings
roller skate bearings will work for my desktop model, but not the big one
hexapod? why is that?
lots of reasons
1 i like to make my own tools, and hexapod is much easier to make than a bridgeport
you'll have a lot of fun homing :)
2 it's portable
3 it looks cool
we'll see about the rest of the claims
supposedly hexapods are much more rigid, and accurate
if you build them right..
if you screw up position of joints, they don't work at all
i've seen some wonky designs out there
* alex_joni wishes you best of luck
i might end up using an icosahedral frame instead of an octahedron
fenn: like this: http://www.mel.nist.gov/photos/jpg/1hexa.jpg
[14:09:14] <alex_joni> http://www.mel.nist.gov/photos/jpg/1hexab.jpg
more or less, but with universal joints instead of spherical bearings
not that big either
have you seen till's "modell"?
that counts as a wonky design in my book :P
probably I've seen it
i also like the ones where the spindle is supported by the 6 axes only and there's no frame
got the link handy? I've misplaced it
[14:14:53] <fenn> http://www.isw.uni-stuttgart.de/personen/t_franit/modell/
[14:16:57] <fenn> http://www.mathworks.com/access/helpdesk/help/toolbox/physmod/mech/mech_cad29.html#24008
that's for workpiece and fixed spindle?
or for spindle, but upside down?
either way really
I mean.. which one will you use?
that particular one looks like it's for positioning telescopes or something
i will go with a frame with the spindle hanging down from the top of the frame
that's the fun of it
workholding is going to be interesting
and emc to control the biest?
emc looks better than anything else out there
probably the hexapod kins in there :D
i was thinking i could do "teleop" mode with brushed servos pretty easy
by constructing a hexapod-style joystick
then the joystick automatically does kinematics for you
good enough for a crane at least
hmm.. won't work on stiff kins
it'll break apart
because of differences between the joystick and the crane?
yup.. and probably because of aliasing errors
well i was thinking all analog
and imprecisions in the joystick
* alex_joni goes home
Hi Ben. How you doing today?
i'm doing alright.. trying to figure out what to do today
Making pieces with that lathe?
that sounds like a good idea
i need to make a better toolpost
I didn't look at the toolpost at all.
4 way or what?
actually it didn't come with a toolpost, so i grabbed one i found on the floor (and it didn't quite work right either)
Okay. The tool post takes a lot of force in a lot of different directions.
i'll be making one like this i think:
[15:34:47] <fenn> http://website.lineone.net/~victoria.ford/gt.htm
The do need to be held in place well.
the compound has this little ridge next to the t-slot clamping surface that gets in the way and doesn't seem to do anything
i want to just chop it off
The tool post in that link looks tough enough.
My only wonder is whether the threads in the alum will hold up over time.
Looking again at the third pic it looks like the tool holder might be steel.
i think i will make a little round nut that sits in a groove so that you aren't bending the screw as you clamp down
That would be a good plan.
A post like that will make tool allignment much quicker.
Say. How do I get a url from logger_aj?
Jacky^afk is now known as Jacky^
anyone tried gthumb ?
ive about 40 photos, need to optimize them for web upload/download..
40 pics maded in sicily island today :P
hi alex_joni !
are you ?
i've about 40 photos, about 1 mb each
you know some app to optimize all for the web ? to upload donwload
about 300 kb each
I use ACDSee
made thumb too ?
but GIMP should be ok
i know gimp ..
ACDSee does it automagically
but using gimp i need to open one at time..
would be nice
there is script-fu
in gimp ?
someone suggest to me gthumb
but i never tried it
gimp has an own scripting language
but gthumb sounds ok
I know about gimp scripts
thanks , going to try it ..
nice photos todayyy :P
yeah gthumb is nicee :P
back from vacation :)
what is vacation ?
not at all holiday .. but 60 %
well, i had 100% holiday :)
i fled from pyopengl
i'm uploading some photos
swig -python -Iinterface -shadow -DAPI_VERSION=137 -o src/interface/GL.ARB._shader_objects.0089.inc interface/GL/ARB/shader_objects.i
interface/GL/ARB/shader_objects.i:73: Error: Syntax error in input.
thats the erroor message that did it for me. after that i straight called a friend who was going to to hollidays and said: im in :)
dammit, still hasnt disapeared by itself
got mesa, glut, tclTk, python ready for puppysit but. that pythonOpengl makes me headaches
yeah, it's a PITA to build
sudo chmod 000 `type -path swig` and start over from a fresh tarball
its swig interface files aren't compatible with modern versions of swig, but it includes pre-built versions of those files so all you need to do is make it think there is no swig available
hi jepler: i got the old swig version. tight now i found out that im missing numpy :) its installing right now
hmm, ok, thats wasnt it
numpy may be required to build pyopengl, but I don't think it's required for axis
we never import it directly, at any rate
pyopengl needs it, says the docs
hmm, ok, stuck again
jepler: without swig it didnt compile either
hmm, i start again from a fresh cvs checkout
thisones 1 week old
the most recent version I've built is 2.0.1.07. I think that's quite old, so all my advice may be wrong.
damn dependency rattail
nice trip with the boat :)
nice day, today
i like this : http://digilander.libero.it/jackydgl0/photos/reggio-messina/img021.jpeg
this is of few minutes ago
[18:55:06] <Jacky^> http://digilander.libero.it/jackydgl0/pics/anmary.jpg
this camera seem work ok
E. 300 :/
src/interface/GL.ARB._shader_objects.0001.inc:1916: warning: `_doc_glInitShaderObjectsARB' defined but not used
cncuser: what is emc2 ?
a warnig should not be a great issue ..
Jacky^: error: command 'gcc' failed with exit status 1
follws the warnings
then maybe youre not using the right gcc ver.
gcc version 3.3.4
should be ok
i used it
well, really got no iea how to get this thing going
im stuck with a python monster unable to be built
dammit i really love to have axis with ouppysit
i'm unable to help you ..
isnt jepler around ?
cncuser: are you talking abous AXIS gui right ?
also got no clues
hmmm, about oythonopengl
puppysit can execute simple pythonprograms and hasa working glxgears. als thats missing is pyopengl
well, i remember when i first tried to compile ardour (2 or 3 years ago) it took me a month to finish :) i dont have the nerves for for something like that anymore
a little of patience ;)
i made allmost a week break, and still no step forward
i cant do more
you cant see anything with this pyopengl, it throws out millions of warnings
[19:12:58] <cncuser> http://
if anyone got a clou, tell me
i really am not going to fiddle around with this shit anymore days. if i dont get it working today, i quit on trieng to get axis on the livecd. i want to finish it. i am not the one that could do the same shit every day, im bored.
ic, the pyopengl stuff seems to be primary developed on windows... counting the carriage returns ..
now im not wondering anymore
* alex_joni just got back
what's the problem?
alex_joni: soirry, couldnt make the upload, still not finished
alex_joni: pyopengl doesnt compile
alex_joni: and it doesnt give me much clues. i tried cvs version and PyOpenGL-2.0.2.01
I wish I could help. Unfortunately, that error message doesn't ring any bells with me.
cncuser: I see only a warning
and then gcc existing
aren't there any other messages?
alex_joni: you seen the box.hinternet.at/pyopenglerror.txt ?
thats the whole story
package init file 'OpenGL/Demo/twburton/__init__.py' not found (or not a regular file)
package init file 'OpenGL/Demo/suite/__init__.py' not found (or not a regular file)
and so on
but i cant find any of these, where should they come from
hmm.. smthg seems fishy
did you run all the steps necessary?
my first impression is smthg like configure didn't run
alex_ joni: well i installed a bunch of pythonstuff
and you're compiling on unfinished data
alex_ joni: there is no configure
python setup.py build_w, python setup.py install
does any of the above complete?
i dont get the python way of install/configure... it looks really sucky to mne
yes the first
so only install is failing?
guess so, at least i dont get any warnings with python setup.py build_w
but the output alsp gives no clou on what has been done
# python setup.py build_w
Togl not to be built change config/linux2.cfg to enable
SWIG Version 1.3.23
Copyright (c) 1995-1998
University of Utah and the Regents of the University of California
Copyright (c) 1998-2004
University of Chicago
Compiled with g++ [i686-pc-linux-gnu]
Please see http://www.swig.org
for reporting bugs and further information
hmm.. afaik Togl is needed.. right jepler ?
why cant axis be built on fltk or something
itas fast, small, and didnt ever made problems when compiling it
but Togl is needed for AXIS
and it has opengl stuff
<cncuser> Togl not to be built change config/linux2.cfg to enable
that doesn't sound right
ok, i can change that
put it on your website
jepler: you remember compiling pyOpenGL?
; a comma separated list of the libs needed when linking Togl^M
; the GL and GLU libraries are included automatically^M
hmm, damn dosusers
dont get it how i can turn something on here
can you post the cfg?
thats the part
theres not much more
which pyOpenGL did you get?
hmm, not shure
Release Notes 2.0.2.01 (alpha) [show only this release]
why dont the just use configure, make and make install
can you try again?
the whole process.. ?
I'll walk it through with you
what python do you have?
2.3 or 2.4?
but i allso can copy the terminalbuffer into a textfile and put it online
Python 2.4.2 (#3, Oct 31 2005, 23:18:49)
ok.. 2.4 sounds ok
did you check the dependencies you need?
[19:34:44] <alex_joni> http://pyopengl.sourceforge.net/documentation/installation.html
yes i have everything + numpy (mentionend in the docs)
GLUT, OpenGL and CLU
alex_joni: I vaguely remember compiling pyopengl.
alex_joni: The only thing I remember is the advice I gave earlier about "chmod 000 swig"
nah.. I think tk8.4 is missing
he only has tcl8.4
alex_joni: axis supplies its own copy of Togl, because at least for awhile the version included in pyopengl was badly broken (segfaulted when you tried to use it)
so don't worry about building the version of togl that is included with pyopengl
jepler: I think tk8.4 is needed, right?
jepler: i ok
anyone a clou where tpo get tk ?
[19:48:11] <cncuser> http://mesh.dl.sourceforge.net/sourceforge/tcl/tk8.4.11-src.tar.gz
Tcl [show only this package]
Release Notes 8.4.11 [show only this release]2005-06-28 13:32
Download tcl8.4.11-html.tar.gz734001 3577Platform-Independent.gz
Download tcl8.4.11-src.tar.gz3474428 27889AnySource .gz
Download tcl8411-src.zip3929559 17498AnySource .zip
ok, tk was missing, got it, now trying with pyopengl-2.0.1.09
* Jacky^ god bless you
cncuser: youre not so lucky ..
im too ..
cat understand wgy 'lost' pixel uploading pics
in local are ok
well, a little
i hate this software 10% more then 5 minutes ago
can you post the log?
its the same
package init file 'OpenGL/Demo/dek/__init__.py' not found (or not a regular file)
package init file 'OpenGL/Demo/dek/OglSurface/__init__.py' not found (or not a regular file)
package init file 'OpenGL/Demo/GLE/__init__.py' not found (or not a regular file)
package init file 'OpenGL/Demo/da/__init__.py' not found (or not a regular file)
package init file 'OpenGL/scripts/__init__.py' not found (or not a regular file)
allways the same
what did you run exactly?
python setup.py build
did you build the other packages?
i dont know what that runs
what package ?
i built mesa, glut, tcl, tk, pynum
whats glu ?
OpenGL 1.1 and GLU
i can run glxgears
so the gl part should be ok
ok.. but also dev part of that?
i compile from source
so of course, i allways have the dev part
else i coukldnt
that doesn't neccessarely mean dev part gets installed into place
it's not enough to build those components
you need to install them
or python foo install
shure i do a make install
everithing is installed
Jymmm: make dissapear
* Jymmm makes cncuser go *POOF*
alex_joni you wanna play like that?
not really .. too tired for that :)
poof was the sound the small nightcrawlers in a xmen comic i read 10 years ago made when the treleport :)
was meant as a joke :P
damn I need like 1/16 cutters =(
is there some other modern looking gui for emc2 ?
axis is it
Jymmm: note the "other" in there
though you could create your own
Jymmm: you name it
its to big for me
alex_joni note the " make your own if you dont like" in there
right.. that's a regular expression :)
i dont know poython, i dont know what that stupid setup.py oinstall routines are for
and i just cant get it to work
also i have fucked up my develimage with dozens of megs of stuff only needed for axis
i go clean
yup.. make it small again
ask cradek or jepler if they can make it work you'll include it
binaries I mean :D
alex_joni: i dont think they have interest in it
they will at one point I think..
and if not.. well, maybe one day puppy-emc will be used intensively, then it'll be worth the trouble
maybe, right now im more or less on my own, and i have a problem with not finishing stuff i intendet to finish within a specific time. i loose all interest, start to hate all parts, thorw it in a dump, and never look at it again.
im aware of that, but i cant do anything against it
cncuser: if it makes you feel better, I also gave up on compiling axis
well, i put the stuff onoline i got when its cleaned. and thats it. nomore puppysit
you are going thru headaches just because of the GUI ?
right.. so skip python, pyopengl, glu, glut, whatever there is more to it..
cncuser: you are going thru headaches just because of the GUI ?
Jymmm: because of some part the gui needs, pyopengl
because it's a nice GUI
Jymmm: like alex_joni says
fenn: its no problem to compile axis on a debiansystem
cncuser: i'm using fedora
redhat's opengl package is screwy
it's been a long time
indeed. I was working on the Grand Challenge for a while, which superceded all else...
cool.. what team?
how precisely did it crash and burn?
* fenn cackles
Very preciesely :)
zwisk: probably you don't know any of the new guys around here..
ok, i out the user iso on my usbstick
gonna go to the next internetputer and uplaod it. cu later
zwisk: some really wacko's around lately
true enough. Seems emc gets a little churn. But that's cool. Keeps it fresh.
I thought I'd try getting axis up and running with emc2. Anybody know if the latest of each work with each other?
some changes happened around here, but nothing too fancy
there's an election for a new board coming up
Yup... saw that.... Pretty cool.
maybe emc will get moving..
Lots has changed on the website in the last many months too... the wiki is great.
yes, the wiki is great
but the website _will_ change a lot
Cool. I think it could use a refresh. Maintenance of websites is one of those things that hardly ever gets done enough, especially in open source projects.
hi zwisk .....
urr.. howdy that should be
very nice day here :P
it is to spanish people
bought sony camera
and maded some nice ptohos this afternoon
take a look to these fish http://digilander.libero.it/jackydgl0/photos/reggio-messina/img002.jpeg
Gonna take pictures of chips flying off a machine? :)
about 50 mt up
what's that photo supposed to be?
I see a lot of blue. You sure there
alex_joni: cant you see the fish ?
are fish in there?
i only see a bunch of sharks
ist there a fish !
the ssmall points you seen are fish :P
Sure those aren't bubbles? :)
no they are definitely sharks
btw solved with the upload 'lost pixel' issue
was binary mode not ascii :/
it seem to me slow ..
alex_joni website is faster :P
.jpg files should be binary
mmhh.. not familiar with it
it probably just messed up and didn't mess up the second time
[20:38:34] <Jacky^> http://digilander.libero.it/jackydgl0/photos/reggio-messina/index2.html
how is ?
it seem to me slow ..
i used gthumb
fast enough for me.. there are visible rectangles around the thumbnails that aren't quite the same as the background
fenn: wich connection ?
how nuch fast ?
im going 1mb download
i dunno.. 150k or so
mmhh.. then should be good
assuming youre outside my contry too
I found alex_joni website very fast
heh.. you did?
not in thumbnails .. i mean
you mean download speed?
loadin full picture is fassst
it should be..
it's on a 100MBit link
you probably don't have enough to stress it out
44878 2005-10-31 18:52 img021.jpeg.medium.jpeg
this the size for medium ..
144789 2005-10-31 18:52 img021.jpeg
this for full
not too much ..
I have some at about 1.2Megs :D
but.. it seem to me slow , from my browser
deleting cache first
computers will suck up all your time if you let them
they are a form of time machine
an excess of time is quickly dissipated into the intricate fins of the device
* fenn goes back to work
yeah.... I'm just tired of always waiting for computers ;)
no matter how fast they get, you still get to wait for them..
yup. The one *true* constant of the universe. :)
as they get fast every 18 months, you'll still wait the same amount of time (every 18 months)
the all new puppysit, with 2.4.31 kernel, even smaller + mail, webbroser, xchat @ puppy 1.0.5
also i changed the usb behavior
as there is muich trouble with usb 2.0 flash
it defaults to use all as usb 1.0
cncuser: hang on, I'll mirror it
21 seconds :D
for 29 mb :)
should I leave the old ctx10b.iso? or only the new one?
please remove the old
i can only support one release
thank you very much :))
almost done ;)
you can put firstname.lastname@example.org as contakt email on your website
i just found out that i forgot to include my email :)
well, the deviliso... hmm, i should have made a backup before starting with the pythonstuff. i try to finish it as soon as possible and tell you when i put it online so you can mirror it
ok.. you can drop me an email..
right now it 500mb :)
but i remmeber it was about 300mb earlier
still quite a bit :)
alex_joni: its the complete puppy 1.0.5 + puppy unleashed + kernel source + emc, rtaissource
the puppy unleashed packages are about 200mb alone
108 downloads on ctx10b
want that to count on?
or a different count for puppysit ?
differnet count please :)
nice, so ctx has around 250 downloads, woow
alex_joi: but i dont need to have access to it. i may ask you every week or so :)
Jacky^ is now known as Jacky^afk
ok, gotta drink some beer, this here is no internetcaffee ;)
cu tomorrow folks
alex_joni: I'm starting to look at the trajectory planner. I'd like to understand it, and I find the fact that it permits infinite jerk to be unesthetic. Also, I seem to still have some blending problems with circles (although it might be that I have configuration or mechanical problems). So... Can you tell me how and where circular (helical) interpolation is done?
not really :/
* alex_joni doesn't know enough about that
Do we know who does?
there's a funny paragraph in the nist g-code interpeter doc...
lerman: not really
paul_c looked at it for some time (not sure how much he grasped)
les knows a bit, not about coding (more about theory)
probably you'll be best if you send an desperate email to FredP
Yeah, I've read and understand the note that les produced. I haven't studied the code that implements it, though. I figure that once I've done that, modifying it to limit the acceleration should be straightforward (use four points instead of three...).
etla: Are you going to tell us what the funny paragraph says???
there are two.
first: The canonical machining functions share with the RS274 language the simplifying assumption
that machine dynamics can be almost ignored. That is, in this model, acceleration and
deceleration do not occur. Components of the machining center can be told to move at a specific
rate, and that rate is imagined as being achieved instantaneously. Stopping is also imagined as
instantaneous. This model obviously does not correspond with reality. The control modes
provided here provide some compensation for this lack of consideration of dynamics.
darn, cncuser is gone?
second one: In CANON_CONTINUOUS mode, the control tries to keep the feed rate constant and does not
try to keep the controlled point exactly on the path at all times. Rather, at junctures between
moves where the direction changes sharply, the corner is rounded. There is a maximum allowable
deviation at such junctures, and the control should never allow that to be exceeded; acceleration
and deceleration may be performed if necessary to do this.
any new ideas?
Currently, there is no function to set the maximum deviation allowable in
so when you switch G64 on, there is absolutely no tolerance for the trajectory planner whatsoever :)
it really shows if you have max_acc set low
etla: how low is low? (not to be confused with how long is a chinese name). :-)
* alex_joni grins
I think I tried with 1/3 of max vel i.e. it takes 3s to accelerate up to G0 speed
also, moves are blended into eachother regardless of angle between them.
Well, that can't be my problem, because I'm cutting full circles and have problems where the half circles intersect. At those points they are tangent to each other and the angle between the line segments is zero. But I still haven't found the code that actually does the circular interpolation. Does it break the circles into small line segments? Or does it do circular interpolation at a...
...lower (closer to the hardware) level?
Jacky^afk is now known as Jacky^
* Yuga slaps Jacky^ around a bit with a large trout
i think circular interpolation is done in the (drumroll) interpolator
not in the trajectory planner
but there may be some kind of blending going on there too..
so who goes first, traj-planner or interpolator ??
"when in doubt, dike it out"
fenn: gee whoda thunk it. Could you tell me which piece of code has the interpolator?
I see sin, cos, atan2 and such in /src/emc/rs274ngc/interp_arc.cc
hold your horses
you mean the cubic one?
that's used after the TP has finished
etla: the TP does circle interpolation
not sure where
probably somewhere inside tc.c
so the input to the TP is restricted to the the canonical lines and arcs
and the exact output format of the TP ? a stream of position commands at the servo loop rate ??
generated by calling tpPos or smthg like htat
I wonder where the different modes G61 G61.1 and G64 are in the TP...
maybe circles are just special cases of cubic interpolation done in cubic.c ?
heh.. think I really need to set you right on this one
* alex_joni opens the sources
/* sets termination condition for motion emcmotDebug->queue */
inside motion control (command.c)
term_cond is used to make the difference between exact stop and feed
so tpSetTermCond() is the one that does it for the TP
I guess :)
but like I said.. I know very little about this *g*
and the motion controller comes after the trajectory planner ?
the motion controller is the one that feeds data to the TP
and expects results
* fenn wonders why this is all so complicated
how would you do it?
well i probably wouldn't call it TermCond for one..
I know, it would be called fenn_s_special_function_beta
like jmka said the other day, I think more clearly defined interfaces and more modular code would be easier to 'get'
yeah, what he said
HAL probably goes a long way towards that for the pid loops, hw interfaces etc.
does somebody have "cncuser"'s e-mail address?
jepler: read back
but above that it seems kind of monolithic
i think some comments at the top of each source file explaining what the stuff in that file does would help make a lot of sense out of it all
maybe more detailed flowcharts also
correct for the current EMC2 debelopment ?
what does tc stand for?
rayh: i never even got started on the toolpost.. *weep*
So many pedistrians so little time.
Has anyone tried remastering puppy -- the emc version?
The remaster script says it doesn't find gcombust and fails.
I was wondering if this was a package they left out for the emc version.
hmm.. not sure
I'll ask when I see him.
Nice simple set of tools on the distro.
does it work ok?
rayh: think the emc-mazak channel is still usefull?
probably not. We can sort though the issues here and wiki.
I can leave it running.. have no problem with that
Then let's do.