ahh wtf.. i locked up a python shell?
pythons dont have shells
thats a tortoise!
and i cant kill it and it doesnt show up on ps??
fenn: you might want to add a quit at the end
that took skill!
wasnt your script
otherwise it left me with python windows all over the place
maybe it was too much for the python to swallow?
i hate the default vtk interactor
I hear they sometimes need a couple of days to digest things
fenn: odd huh?
pastebin.ca is faster btw
fenn: can you explain why it does this?
[01:04:02] <fenn> http://pastebin.ca/323105
i gave up on pastebin.com
"maybe it was too much for the python to swallow?"
great line, I shall use it someday :)
first of all.. why is your cone a cube?
fenn: haha, I stole a example and didnt bother to change it..
then I noticed the problem with the cubes..
the python shall bite your a$$ for that
03cradek 07v2_1_branch * 10emc2/debian/extras-Ubuntu-5.10/emc2.files: finish adding gcoderef to the cnc menu, take the udev restart out of -sim
with the z offset..
03cradek 07v2_1_branch * 10emc2/debian/extras-Ubuntu-6.06/emc2.files: finish adding gcoderef to the cnc menu, take the udev restart out of -sim
or whatever you would refer to it as
anonimasu: h/2 - 0.5
fenn: but that breaks it on even stuff ;)
nope it works
[01:26:12] <fenn> http://pastebin.ca/323127
try with H = 5
i even tried zero and various decimal values
the cube on the left must be at -0.5
that solves itself..
as all cubes will be generated at the plane..
then /2 should fix it..
although GetCenter says (0,0,0)
gee that's weird
I did offset the first cube to 0.5 and it works..
it works then..
but it seems like a bug..
it's the offset of the height..
how stupid of me
you always need to do H/2
to get the cube at the same plane..
if the cube height is 1 the origin needs to be at 0.5 for the plane to match :)
once you make it a proper function you wont ever see it..
that's what i'm doing now
im writing a stick array..
how large do you think sticks should be?
or well, ofcourse it has to be dynamic..
[01:41:28] <anonimasu> http://code.enthought.com/ets/doc/api_docs/enthought.tvtk.array_handler.html
03cradek 07v2_1_branch * 10emc2/tcl/tkemc.tcl: backport 1.41: experimental fix for problem with abort offset confusion
hm, apparently "Numeric" is what to use for arrays..
it's faster then the built in lists..
i made a matchstick class
you are faster then I am
im working on getting a 2d array..
if you show me yours i'll show you mine
I dont have lots yet
from Numeric import zeros
A = zeros( (100,100) )
what does zeros do?
a 100x100 array of zero?
[01:52:47] <anonimasu> http://www.penzilla.net/tutorials/python/numeric/
fenn: "empty" works also
instead os zeros
and it's 200% faster..
that's probably what we want..
and then size(A)
arrays in python seems rather messy
* fenn decides not to ask his question and take the easy way out
from Numeric import *
A = empty( (10,10) )
for i in xrange(0,size(A)):
Hm, not right.
needs another dimension.
python is really not nice with arrays.
ok.. what's goin on here
self.add(1, 1, 1)
TypeError: add() takes exactly 3 arguments (4 given)
remove your spaces
fenn: the 4th argument is the implicit 'self' argument
or, rather, the first argument is -- the third 1 is the 4th argument
does that make any sense?
so i should have def add(self, x, y, z) or whatever
03cradek 07v2_1_branch * 10emc2/app-defaults/XEmc: font fix
03cradek 07v2_1_branch * 10emc2/scripts/emc.in: starts too fast - temporary hack
from Numeric import *
A = empty( (length,height) )
for yI in xrange(1,height):
for xI in xrange(1,length):
there's an easier way to set all the elements to a scalar
>>> n = Numeric.empty((3,3))
>>> n[:] = 7
i have an array of cubes
its horribly wasteful i'm sure
fenn: share your code :D
good night all
see you alex
fenn: I made a sticks class now
i think i forgot to set the z position somehow
from Numeric import *
stickmatrix = empty((height,length))
creatematrix(10,10,10) # Height in mm, Width in mm, resolution per mm
ok so how do you add an object to the array?
you have an array of numbers
fenn: x/y will be the cutter coordinates..
i have an array of objects with no numbers associated with them
well, actually ou have an array of empties
fenn: I'm going to set it in another function..
so you can create the array then set the height for everything
or should I make it set height too?
wait a sec.
uh oh my brain just gave out
coding from scratch is so much more fun than trying to decipher someone else's code
from Numeric import *
stickmatrix = empty((sizeY,sizeX))
stickmatrix[:] = defaultZ;
creatematrix(10,10,5,10) # Height in mm, Width in mm, resolution per mm
there you go
shouldnt that be sizeX*resolution etc
ok so.. we have this array
oh i get it
i'm just displaying the array
you need to call the stick array with the same arguments
to create the offsets for all stocks
the x/y offsets..
uh, i dont like the global
we should make it a class..
Or add it into your class..
will try to work it into what i've got for heightmap()
anonimasu, what time is it?
fenn? what was that about LH being my henchman?
fenn: that way we can grab the offsets at the same time..
fenn: this is wrong..
fenn: matchsticks should go into a class called stickgrid.. or something
err heightmap ;)
my head's messing
self self self self self
fenn: I think this stuff will be nice when done
self.add(i,j,A[i][j]) #something pretty
my head just ceased working
A[i:j] gives me [5 5 5 5 5 5 5 5 5 ... 5]
ok a[i][j] works
ok i think it works
paste your code :)
its very slow with 10 cubes per mm
can it be your pc?
[02:52:30] <fenn> http://pastebin.ca/323193
runs super smooth
eh, that's not the right one is it?
that's only 200 cubes
apparently 200 is ok but 1000 is too many
maybe it would be better to use vtkQuadPoly or something
hm, yeah maybe not a proper box..
I were thinking of using pyopengl and just creating the 4 sides manually
in the window..
hm, the performance is pretty bad..
that's just 800 polys..
if it does render all sides..
let's think about this more tomorrow :)
aww you wimp
i'll poke around and see what i can do
yeah do that
we'll find a way somehow :)
fenn, cool. I got it running here- your code
fenn: it shouldnt be slow.
[03:25:18] <anonimasu> http://rrvtk.free.fr/vtk/perfs/
03cradek 07v2_1_branch * 10emc2/configs/sim/tripod.ini: this halfile is gone
03cradek 07TRUNK * 10emc2/configs/sim/tripod.ini: this halfile is gone
03cradek 07TRUNK * 10emc2/configs/common/emc.nml: this breaks 32 bit machines!
cradek: why did you take that out again? is there a new report that it broke something?
it broke my machine
I guess that by running 2.1 and 2.2 concurrently you'll be running into the "leaves shared memory segments" bug a lot
but that's the problem, not the nml file
I forgot that was the problem
after ipcrm -M <some number> 2.2 will run just fine with that nml file
at any rate I believe that's what's going on
I was pissed while dealing with other breakage, I should have looked into it more first
sorry about that
nah, my fault
I'm watching a particularly bad episode of star trek: deep space 9 and it's got me on edge
wow I have a lot of shms around
the 3xx ones are emc, I think
which particularly bad episode?
skullworks: odo gets an infant changeling, and kira finally gives birth
* skullworks wonders where he put that DS9 complete box set.
03cradek 07TRUNK * 10emc2/configs/common/emc.nml: oops
any idea what may be wrong with the supplied lathe similation config? It complains about a signal already having an output pin. This is the checked out config and nothing obvious from inspecting lathe.hal
[04:43:13] <ds3> http://www.pastebin.ca/323246
ds3: they are working on that now.
bull-dozin berry is back.
where was it?
out and about.. good mouser.
likes to head butt
i need a one of those for all rodents, not just mice
only problem is most cats cause an allegeric reaction :(
that sucks. (well if you like catsO
yes, I do like cats. they are useful
the best arrangement was a stray that visited/lived in the yard. but then neighbors got dogs :/ no more cats, just barefruit tree. die squirrels die
:) all of our cats are strays..
me and my wife..
berry make it half way up one of our cherry trees when a dog came around
the disease-carrying capabilities of strays don't bother yoiu?
they where strays. now they are 'fixes' and up to date on thier shots.
bug reports everywhere. haha.
can anyone help? `diff \"Step NC-code\" \"G-code\"`
need a \ after Step
`diff Step\ NC-code G-code`
maybe a -R too
no no, I was being nerdy
what's the diff b/n step-nc code and g-code? I thought g-code was aka nc code...
that would be a good subject for a book
the long and short of it is that g-code is a low level data format and step-nc is a high level data format
G-code says "go this way, this fast"
STEP says "make this shape, taking into account the machine capabilities and material properties"
so they're a little different :)
step-nc has an internal version of g-code too, i think
it has to generate the toolpath some time
03cradek 07TRUNK * 10emc2/src/emc/motion/ (control.c mot_priv.h motion.c motion.h): fix lathe threading
03cradek 07TRUNK * 10emc2/src/emc/kinematics/tp.c: fix lathe threading
03cradek 07TRUNK * 10emc2/configs/sim/lathe.hal: fix lathe threading
time to get drunk and go play FEAR - nite all
03cradek 07v2_1_branch * 10emc2/src/emc/motion/ (control.c mot_priv.h motion.c motion.h): fix lathe threading
03cradek 07v2_1_branch * 10emc2/src/emc/kinematics/tp.c: fix lathe threading
03cradek 07v2_1_branch * 10emc2/configs/sim/lathe.hal: fix lathe threading
sounds like a cvs update is needed for me
and bedtime is needed for me
03jmkasunich 07TRUNK * 10emc2/docs/man/man9/ (freqgen.9 stepgen.9): add man pages for stepgen and freqgen
03jmkasunich 07v2_1_branch * 10emc2/docs/man/man9/ (freqgen.9 stepgen.9): backport: man pages for freqgen and stepgen
no no no, I'm most definetly male :p
just a typo
anon here's the latest http://pastebin.ca/323410
how's the simulator coming?
and it doesnt simulate anything yet
hey, it's something
how slow is slow?
not bad until you hit a certain threshold
oh, I see
and is that image above or below it?
that's only 20x20 or so, so that shouldn't be hard at all
yeah that's plenty fast
what about setting the matchstick size to 0.5x0.5 mm and having a part say 200mm x 200mm
so 400x400 elements
so 100 times as many elements
well, it doesnt work, i mean
what's the bottleneck?
feel free to fix it :P
alpha's here, guess its time for me to go to bed
I just got in
some asshole wouldn't leave my friend's party... so I stuck with my friend to make sure the dick left.
called us assholes cause we didn't know who the fuck he was, and the party was over, so we got him to wake up, and asked to call him a cab.
guy spent 30 minutes trying to call friends to see if he could crash there.
no one wanted to pick up the phone for him... and eventually called this girl he knows is mother...
to go to the girl's house.
WTF... what a loser.
SWPadnos_ is now known as SWPadnos
SWPadnos, SWPadnos, SWPadnos, he's our man... if he can't... so many maybe can.
emc development, where the sun never sets!
03alex_joni 07TRUNK * 10emc2/docs/UPDATING: put info on configs into TRUNK too
im looking at how to get more performance with the vt
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-01-21.txt
03alex_joni 07v2_1_branch * 10emc2/src/Makefile: install UPDATING doc
anonimasu, how bad is it now?
lerneaen_hydra: very bad
is that becuase of non-hardware 3d drivers or something else?
lerneaen_hydra: no, the practice we are using
you mean the entire matchstick model?
you dont need a separate actor for each stick..
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-6.06/emc2.files: include Integrator Manual in the menu, provide sane comments to the menu items
I am sorry to bother
I am looking for alex
anonimasu, oh, I see
anonimasu, are you using a zbuffer?
03alex_joni 07v2_1_branch * 10emc2/debian/extras-Ubuntu-5.10/emc2.files: include Integrator manual in the menu, provide sane comments for the menu items
03alex_joni 07v2_1_branch * 10emc2/docs/src/ (Master_Developer.lyx Master_Integrator.lyx): bump manual versions to v2.1
03alex_joni 07v2_1_branch * 10emc2/docs/src/install/installing_emc2.lyx: no more emc2-axis for 2.1.x+
03alex_joni 07v2_1_branch * 10emc2/docs/src/install/compiling_emc2.lyx: we refered to HEAD, but meant TRUNK
lerneaen_hydra: Z buffer
lerneaen_hydra: that's not related..
lerneaen_hydra: this is with vtk whatever method it uses.
well, with something like matchsticks not having a zbuffer will cuase really bad performance
lerneaen_hydra: that's not the ussue..
lerneaen_hydra: the issue is that we are creating 3000actors..
lerneaen_hydra: vtk should be fast even with lots of polys..
darn, the manuals have really grown
lerneaen_hydra: apparently VTK should be blazingly fast..
as it's a very thin wrapper
anonimasu, oh, so there's not much overhead in it
alex_joni, isn't that a good thing? ;)
lerneaen_hydra: sure is.. but makes proofreading a pita
3x200 pages or so
btw, are there any places where you need images/graphics?
how do you mean that?
well, is there some place where you know you would like an image/graphic to clarify something
is there a list somewhere?
or just in your head?
in the beginning of the user manual there are some ugly older images
first couple of pages
the developers manual is really lacking
especially some graphs about system overview
hmm, I'll take a look at that
but it's a problem to draw those if you're not familiar with the system
is cvs the simplest?
oh, I see
graphics: quite a few images/screenshots need updating
the config picker, halmeter, halscope, etc
ah, I see
the tutorials would need someone to run the through again, and check for changed commands/names
lerneaen_hydra: plenty :P
sometimes emc2 evolves so fast, it's not possible to keep the docs in sync
docs meaning: lyx, man-pages, wiki, etc
and writing docs is really an unrewarding job
why comment when you can code :D
with code you see that what you do is usefull
hence, the commentator: http://www.cenqua.com/commentator/
people use it, argue about it, improve it, whatever
03alex_joni 07v2_1_branch * 10emc2/docs/src/Master_User.lyx: fix alignment of config/stepper.lyx
lerneaen_hydra: page 11 User Manual
Figure 1.7: EMC2 Process Diagram
looks like sh*t
the file is called whatpc.png
[14:39:12] <alex_joni> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/~checkout~/emc2/docs/src/common/whatpc.png?rev=1.1;content-type=image%2Fpng
Not only does it look like sh*t; it doesn't say anything. What do the arrows mean? Let's see... the g-code is input to the GUI (note that Axis isn't listed). Well, not really. The g-code is input to the interpreter... but that isn't shown. The leadscrews output to the machine axis. No. That isn't right. They move the machine axis. The drawing doesn't mean anything. Or maybe it means everything.
A drawing doesn't mean anything unless the notation is well defined. In some case that is obvious (the meaning of the notation). In this cas it is NOT.
It does say overly simplified - that part is right. ;)
I would show: g-code input to interpreter which feeds canon (motion control) which feeds HAL which feeds drivers which feeds hardware. I would show gui probably with a dotted line connected to (I don't know where that connects -- canon, hal?).
Also gui can feed interp (MDI case).
Do we have a standard for figures? I would suggest that figures be in some standard format (or be convertable to that format). Say: .PNG files with in 8 bit grayscale with a size of 8 x 10 and a resolution of 300x300 dots per inch. They should not have figure numbers or captions as part of the image.
um, IMO graphics should be a vector format
Anything more restrictive than that would make it harder to have multiple contributors.
easuer to edit later on and no rasterisation nastyness
LH: I can't disagree, but is there enough choice on tools to generate it?
there's plenty of OSS apps available that use it
and it's quite a common format
I recently used both Visio and Alibre to generate figs for a patent application.
maybe not in MS paint, but people that use MSpaint aren't doing to make graphics for things like this
I would NOT like to have to learn a new drawing package.
they probably can export to svg if they're vector based
True. I suggested a bitmap format because everything can convert to it.
how about something like vector format preferred, else png, 300dpi, etc
Visio gan generate dxf, dwg, igs, ps, eps, and a bunch of bitmap formats.
no svg O_o
oh, it's a microsoft app
Alibre can also generate a large variety (I pick those two because I'm familiar with them ).
I would say a max size of 8x10, smaller is fine.
True no svg.
ps is vector isn't it?
8x10 isn't a size as such
pixels are size ;)
I suggested 8x10 because you can always scale down. But scaling up can look crappy for bitmap formats.
Yes ps is vector. Is there a ps to svg converter?
8x10 what? inches? at what resolution?
I suggested 8x10 inches at 300 dpi (or should that be dots per linerar inch).
so 2400x3000 pixels
sounds about right
lerneaen_hydra: notebook paper size
Multi-headed LH's use some very large notebooks. :-)
indeed they do
standard ones look very punitive
A2 paper sheets are my preferred size
Not 'B' size?
googled ps to svg. There appears to be a converter. Status isn't obvious.
IMO svg is a good format because it's open and XML based, so backwards compatability is not an issue
[15:09:42] <lerneaen_hydra> http://en.wikipedia.org/wiki/Svg#Example
pdf2svg released: www.svg.org/story/2006/3/23/14251/5858 - looking now.
if you make all images 2400x3000 you'll have quite a bit of weight on the CVS checkout
What tools would you use to generate it? On Linux? On Windows? Free?
IMO inkscape is good
openoffce draw can export to it
gimp can too
and cross platform
inkscape is really nice: http://en.wikipedia.org/wiki/InkScape
(the missing svg bits are not very important for the type of things EMC needs)
Never mind - the link I sent was not OSS
if you're running KDE then karbon-14 seems nice
Inkscape sounds good for 2D stuff. What about 3D stuff? (Or don't we need it?)
In the win environment I use visio - for linux I'm still using paper and pencil.
I didn't know that linux supported paper and pencil :-)
why would we need 3d stuff?
and what would the type of modeling be then?
I've been using quill and parchment. :-)
that does make it difficult. I'm open for suggestions for linux cad.
If you wanted to show a (2D projection of) milling machine with the coordinate system.
Or a hexkins.
or just draw it in psuedo-3d
That's OK for an artist. But not for an engineer (like me).
(pseudo 3D, that is)
if it's just a projection to show the X/Y axis of a mill, why not?
Hard to change. Gee, I can't see that feature real well. Could you rotate it another 20 degrees? Sure (I'll just have to redraw it and change most of the lines).
yeah, that's true
you'll have to make up your mind first and decide exactly the layout
take a look at blender if you want an OSS solution
Even if it were just me, making up my mind would be hard. Satifying even 50% of us here would require multiple iterations.
how often though is a real 3d model needed
But I could cheat. Keep the original in Alibre. Export to postscript and convert (if it works) to SVG. Import to Inkscape. Pretend I had done it all in Inkscape. When you want a change, smile and say "Sure, nothing to it."
Ideally, though, "we" would own the source (the original).
lerman: send me the alibre file(s)
I don't have any (for this project) yet. It's all hypothetical.-- so far.
as long as the format isn't bastardised when converting (merge to one layer, kill gouping, etc) then I don't see why not
if it makes a semi-rasterized mess of it though, it would be hard to edit the file
lerman: I know..
On another topic... I was talking to skullworks (I think) yesterday about macro libraries. It seems like I could have the interpreter automatically import them. When it sees an Oword (say Onterpolatecircle) it doesn't know, it would look in some fixed directory for a file called 'interpolatecircle.ngc' and use it.
Or whatever. O words will soon have names rather than just numbers.
I take it you mean Ointerpolatecircle
presentation tools for linux, scalable, svg, pdf, ps, & free see scribus http://www.scribus.net/
and sample of emc doc http://imagebin.org/6976
or is this limited to macros starting with i ?
IMO an underscore might be nice, say O_interpolatecircle
Well, the actual syntax would be O<interpolatecircle>.
you could call your files _interpolatecircle though, right?
We could call them whatever we specified.
ok, sounds good
That's the least of the problem.
issues: (1) how do we specify the directory to look in (2) is there more than one such directory (3) what does Axis do with this.
Axis (and other guis) would need to be able to see the current file name as well as the current line number. I don't know how difficult that would be.
I think it would be really easy to implement.
there's already a mechanism for M codes, isn't there?
Yes. I don't know how that works. What directory does it look in?
not sure or no clue
This is another 'no brainer' though. Anyone who doesn't like it doesn't have to use it. And it won't break any existing gcode.
gotta run - checkout time is approaching and Cabin Fever beckons.
re OinterpolateCircle, so my OinterploateCircle is used when in the source, else the library OinterpolateCircle is used when not in the source
it can be overloaded
hey people. If I'm running tkemc, do you think there's another application that I could run at the same time, for display the tool path of my machine?
Exactly. If it hasn't already seen it, the interpreter goes and finds it.
er, "display OF the..."
owhite: besides backplot? in addition to backplot?
tomp_: pardon my ignorance -- does backplot display the position realtime?
owhite: you found mine ( ignorance): i dont recall if it's a dro, but it does "display the tool path"
owhite: What do you mean by realtime?
I was hoping to find an application that will do two things a) show the overall part that will be cut, and b) show the position of the cutting tool while the machine is performing the run.
ah. backplot shows the "position of the cutting tool while the machine is performing the run."
okay. so if I typically do the run in tkemc, can I also run backplot at the same time? do they communicate with each other?
Yes they do
* owhite knows he's heading dangerously close to "rtfm" land. :-).
backplot is under the view menu.
The only issue I know of with backplot is running your program so many times without clearing the backplot canvas that you run out of memory.
I heard of a Sherline user that did that.
rayh: okay. welll I'll look at it. Apparently I've broken my tkemc script when I use that menu item :-)
rayh: while I have you here though, I have another question with the USC board.
so you've recently added code to support the additional IO. but one of the things I realized is that I really would like to control some items when the machine is in e-stop.
so for example. I was hoping to control my laser's power supply, to turn off the overall laser system (not to fire the beam).
but unless the overall laser is powered up, it will be in e-stop.
with things like lube, coolant, and such you may run up against EMC's hard coded logic.
yeah. that's the problem.
You could turn on an individual pin with something like halvcp -- pyvcp
that would be great, could I do that in tkemc if I hacked it a little?
You could also "or" that signal with the one from EMC
I've not tried to get access to HAL stuff from tkemc but it should be possible.
If you know a bit of tickle you could look at halshow and find the relevant code to toggle a pin.
basicly we use halcmd
so I'm googling emc2 and halvcp and not finding much.
and send a "sets" command to the signal you create for the control
You won't. Most of the documentation is in the EMC documents.
What version of emc2 are you running?
* owhite wouldnt set foot in the forum unless he was running emc2-head. :-)
The head docs are available at linuxcnc.org
I know the feeling.
2.1 will really close that gap.
okay. I'm happy to look into it. poke around for halvcp, and use that to drive the pins?
owhite: to look at halcmd, you look at the src. docs at that level >are< the src, not any handbook
You can test your commands using halshow
What are you using for IO
owhite: it's called trunk
Thanks. Hi alex_joni
HEAD is the latest version
but that works for a branch too
TRUNK is the main branch
and of course it has a HEAD.. but so do v2_0_branch and v2_1_branch
rayh: what do you mean? I am using the USC, and then wiring to the new port (I forget its name) and then the pins on the USC will drive a relay board.
so using only HEAD is confusing
* owhite is using TRUNK.
trunk trunk trunk.
tomp makes elephant noises
Um. I know very little about that addition to USC
owhite: what hardware do you wire to on the computer side? parport, stg, some card?
but you can create "newsig manuallaserpower bit"
* owhite confused rayh with jon elson.
duh. my mistake. different board member.
no problems. I know Jon well and have a USC card.
ah. well he added the capability to use P8.
next you should "linksp manuallaserpower usc.xxx.yyy" where you supply a good name for the pin you want to use to turn on power to the laser.
I still wonder if since the board's IO is tied to e-stop if I will be able to do this, but I'll poke around.
Ah. You may well be right.
His estop does shut down the ability to pull his output pins.
You might have to add a second parport or a DIO card.
I figured. do you guys know of a way to drive 8 or so bits out of the USB or serial?
actually, parport cards are pretty freaking cheap.
Yes they are.
are there any problems with tkemc, or emc2, for driving two parellel ports at the same time?
No. I've done it here quite often.
You just have to edit the relevant .hal file for the extra port.
so cool. any documentation or examples anywhere?
* owhite awaits the dreaded "are you kidding?"
* rayh goes to look at the usc config
alex_joni: nows your chance to express the usual "wtfm".
it seems to me that univstep_load.hal would be the place to add the pci parport.
okay I'll do some research.
i didnt find 'use 2 parports' but did find 'use 1 parport for somethng besides xyz stepper' http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Torch_Height_Control
owhite: loadrt hal_parport cfg="0x378 0xfb00"
that will load 2 parport drivers
alex_joni: which hal file would that go into?
remember that one parport is already taken up with usc
the torch height example does load 2 parports http://wiki.linuxcnc.org/uploads/dallur-advanced.hal
yeah hopefully something funny wont happen to the USC if I run two parports.
Are you running the usc from the mobo parport?
ah. yes. the motherboard.
I'm just a bit confused about how hal_ppmc knows the parport address.
I presume that it is just polling until if finds the card.
is there a default? some of the config directories dont seem to specify it.
I added a pci parallel port card here and found it's address at 0xc000.
so the command to start a hal for it is "loadrt hal_parport cfg=0xc000"
* owhite goes back to doing laser stuff.
sharks! and lasers!
* owhite just built a new ventilation hood, and mounted a little rubber shark to it. :-)
if I cant have a sharked with a mounted laser, at least I can have a laser with a mounted shark.
03jepler 07v2_1_branch * 10emc2/debian/configure: by changing to the directory of the script, running 'debian/configure sim' from the top level works
has anyone tested classicladder in these new 2.1~alpha0 packages?
03jepler 07TRUNK * 10emc2/src/hal/drivers/hal_ax5214h.c: trivial grammar fix in comment
03jepler 07TRUNK * 10emc2/src/ (Makefile Makefile.inc.in): correct version of the driver for the futurlec 8255 card
03jepler 07TRUNK * 10emc2/src/hal/drivers/ (pci_8255.c hal_8255.c): correct version of the driver for the futurlec 8255 card
ooh the freqgen/stepgen manual pages
inkscape is really nice
alex_joni: did you commit the changelog for ~alpha1? I still have ~alpha0 in my debian/changelog
let me see
I made me a dragon. nilno.com/head1.jpg
03cradek 07v2_1_branch * 10emc2/configs/dallur-thc/dallur-advanced.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/demo_mazak/demo_mazak.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/demo_sim_cl/demo_sim_cl.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/demo_step_cl/demo_step_cl.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/etch-servo/etch.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/halui_halvcp/halui.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/m5i20/m5i20.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/lathe-pluto/lathe-pluto.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/hexapod-sim/minitetra.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/max/max.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/motenc/motenc.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/nist-lathe/inch.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/ppmc/ppmc.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/sim/ (8 files): cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/stepper/ (sim_inch.ini stepper_inch.ini stepper_mm.ini): cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/stepper-xyza/inch.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/univpwm/univpwm.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/stg/stg.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/univstep/univstep.ini: cleanup for release
03cradek 07v2_1_branch * 10emc2/configs/vti/vti.ini: cleanup for release
just for giggles - Whats a lathe-pluto?
pluto is an FPGA interface that generates PWM and counts encoders
hal driver for a pluto-p running a lathe?
more about the pluto servo driver: http://linuxcnc.org/docs/2.1/html/hal/pluto_servo.html
* lerneaen_hydra pokes A-L-P-H-A
is PM'ing broken?
are you "identified"?
you know: /msg nickserv identify <password>
A-L-P-H-A, are you ident-ed?
pluto looks useful
yez - JTAG
same fpga on a pci card, bit of buffering ...
robin_sz: I like the mesa better - much bigger FPGA on a PCI card
still no buffering tho
is there a mesa driver yet?
I was unaware of pluto
there is a driver for the standard 4 channel mesa config
and for the pluto servo
the mesa card is big $$$ though right
pluto is $trivial
less in quantity
a group of us have purchased mesa's and once version 2.1 is out we're gonna be working on additional configs and drivers for it
but you need the buffer cards and expansions etc for the mesa to make it useful though
which crought it to nearer $400 when I checked last
depending on what you want to do, yes
robin_sz: I think 300$ is more like it
depends how "cheap" you are though
for $60 ... id bess about with optos etc :)
* robin_sz is cheap
I mean what you want to built yourself
robin: and what you want to fit in the FPGA
I'm not sure how many gates are in the pluto FPGA
10K, 20K, something like that
the mesa is 200K
not caring ... if it can do 3 channles of servo thats enough
plus soem digitial IO
the pluto is probably for you
I presume on the Mesa you could put the servo loop in the fpga too
alex_joni: when i did "dpkg-scanpackages binary /dev/null | gzip -9c > binary/Packages?.gz", i got a file named 'Packages?.gz' :)
did I write that?
oh.. pesky wiki :)
(yes) should i strike the '?'
the wiki software adds that automatically
I fixed it.. thanks
thank you for making it simple
here's an idea..
one could use CD's or DVD's with lightscribe support as cheap encoders
they have very fine markings on the inside (probably for syncing the image burned on it)
that might make a great spindle tach
thos US robotics encoders are only $19.95
maybe ' sudo apt-get update && upgrade' should be ' sudo apt-get update; apt-get upgrade' , & i like the encoder writing idea, even grey encoders, and THRU HOLE too!
$19.95 in quantity - and no Index channel
tomp_: yes, I assumed people who will be trying this know that
tomp_: don't want to make it too easy, so as not to encourage less advanced users to use it
I just did a label for emc2 disks using lightscribe
pictures coming in a few minutes
I made some encoders for someone out of sheet metal.
owhite: easier with a laser
well if people want something simple I could zip some up.
it probably had like 128 bit resolution.
128 cpr you mean
the chip on the pluto is about 1/20 the size of the one on the mesa (10k vs 200k gates)
are you guys talking about the boards at knlm.com?
[18:32:45] <alex_joni> http://dsplabs.cs.utt.ro/~juve/dropbox/lightscribe/
alex_joni, did you do that running *nix?
lerneaen_hydra: unfortunately no
that's what I was afraid of
the linux lightscribe package is somehow crippled as far as I'm understanding it
there is something, but not fully OSS
acemi is now known as acemi_
acemi_ is now known as acemi
and on doze I had to install an aditional thingie to increase contrast at the expense of time
that one took about 25 minutes
wait, are those burned into the CD by the laser?
which is just wrong imo
Alex - those look sharp!
jmkasunich, yeah, with the laser
and it doesn't affect the data at all?
jmkasunich: burn a full DVD in 5min
flip it over, burn the label in 25
my drive has a competing tech, it's based on blue colors though
lerneaen_hydra: lightscribe has 8 colours available
alex_joni: looks very nice. are the central marks your's? ( what looks like encoder markings already )
no, those are on the blank
when I did labelflash I got good contrast at around 10 minutes burn time
so the software can sync the writing (my guess)
lerneaen_hydra: depends on the coverage
I had a full picture (on the whole disc) and it took 30 minutes for crappy contrast
oh, with mine full image took 15 or so
oh, I see - special disks and drives
* jmkasunich loses interest
jmkasunich: for now..
lerneaen_hydra: not that expensive
5-10$ more for the drive
hey, it occurs to me, I have something that could burn CDs. :-)
here in sweden they're 4x the cost of standard disks
blanks are 20%
the drive is about as expensive here
yeah the media does command premium prices
20%-50% more pricey (but I hope it'll drop)
a cool solution would be milling into the disc for DVDs
not really cool
as the data layer is in the middle of the disk
it'll ruin your drive if you don't mill symetrically
hmm, I'm not too sure, i've seen several disks that are offcentered by nearly a mm or so
on DVD's even a sticker/label glued off-center can make the disk unusable
lerneaen_hydra: not in all drives..
okay, use CAD that does autobalancing ;)
or get it to a tire checking service afterwards
to glue little lead thingies to it
M$ got sued over a batch of W95 CD that had heavy silkscreening on half the disc - and on slow PC's the drive was in use - out of balance long enough to cause drive failures during an install.
heh.. and that was at low RPM
nowadays it's quite high :)
18x on DVD's
that's what my drive does
funny, not a 2^n speed
then again, neither is 12x
no, it's 16x with C<mumble>
linear or angular ;)
probably some psuedo mix
yeah.. something like that
CAV on 18x, ZCLV on 16x
does CLV up to 16x, gets the last data with a CAV stragegy and gets slightly higher speeds there on the outermost data
I think they also use multi track reads - like a download accellerator to increase transfer rate
or same laser but different tracks?
03alex_joni 07v2_1_branch * 10emc2/debian/changelog: bump to alpha1
fuzzy lasers :D
single laser - dual pickups
I used to work for GTW - but got outsourced along with almost all of the U.S. staff
dealt with warranty related failures...
pretty busy then huh?
03jepler 07TRUNK * 10emc2/tcl/bin/pickconfig.tcl: having a writable, personal nc_files directory helps a beginning user.
03alex_joni 07v2_1_branch * 10emc2/debian/configure: print distrib name
kinda - shipping product before documentation was out - shipping without drivers...
Window ME (Microsoft Experimental) was a total fiasco.
maybe vista will be the same
what, MS changed their strategy from "it compiles, ship it" to "it should compile, ship it"?
was nice to go home and have a nice stable network I could use
Vista will be terrible for hardware!
no more unified drivers
IMO vista has so much DRM crap in it that even normal people will get pissed
and blueray has allready been cracked in xp
there are vids of it on Youtube
all drm can be cracked given time
problem is M$ is now dictating hardware specs... not the other way around
so even when you don't use M$ your stuck in there shadow.
awalli1 is now known as awallin-VM
03awallin 07v2_1_branch * 10emc2/docs/src/hal/vcp.lyx: add pyvcp documentation to 2.1 branch
03awallin 07v2_1_branch * 10emc2/docs/src/Master_Integrator.lyx: include vcp documentation in the 2.1 integrator handbook
03awallin 07v2_1_branch * 10emc2/docs/src/hal/vcp.lyx: clarify pyVCP/VCP status
03alex_joni 07v2_1_branch * 10emc2/configs/stepper/README: update with sim_inch.ini config
03jepler 07TRUNK * 10emc2/src/hal/classicladder/arrays.c: don't pop this message up like it was an error
03jepler 07v2_1_branch * 10emc2/docs/src/Submakefile: add vcp html documentation
03cradek 07v2_1_branch * 10emc2/src/hal/classicladder/arrays.c: don't pop this message up like it was an error
03alex_joni 07v2_1_branch * 10emc2/configs/dallur-thc/README: change wording a bit
03alex_joni 07v2_1_branch * 10emc2/configs/dallur-thc/dallur-advanced.ini: G50 is now M52 P1 / M52 P0
03alex_joni 07v2_1_branch * 10emc2/configs/demo_mazak/demo_mazak.ini: G50 is now M52 P1 / M52 P0
03jepler 07v2_1_branch * 10emc2/docs/src/hal/ (16 files): merge from HEAD: images for pyvcp documentation
03alex_joni 07v2_1_branch * 10emc2/configs/halui_halvcp/README: mention pyvcp
03jepler 07TRUNK * 10emc2/docs/html/gcode.html: document new codes m50..m53; remove documentation of m48/m49 since m50/m51 do the same job better
03alex_joni 07v2_1_branch * 10emc2/src/Makefile: fix genhexkins by including _posemath and sincos
03cradek 07v2_1_branch * 10emc2/tcl/mini.tcl: read program dir from ini
03alex_joni 07TRUNK * 10emc2/src/Makefile: fix genhexkins by including _posemath and sincos
03cradek 07TRUNK * 10emc2/tcl/mini.tcl: read program dir from ini
long time no see
03alex_joni 07v2_1_branch * 10emc2/docs/html/gcode.html: merge revision 1.2 info about FO (feed override), SO (spindle override), AF (adaptive feed) and FH (feedhold)
adptive feed hold
adaptive feed control
0..1 as a hal float pin
adapts to something like EDM voltage
elson: I'm building a powersupply, but I'll be testing your drives real soon...
yeah I know..
03alex_joni 07TRUNK * 10emc2/docs/AUTHORS: add awallin
03alex_joni 07v2_1_branch * 10emc2/docs/AUTHORS: add awallin
alex_joni: that's very kind of you ;)
awallin-VM: did you see the test fenn did of the matchstick stuff?
awallin-VM: if I saw you didn't do it yourself :P
awallin-VM: not nearly fast enough with vtk
awallin-VM: but that's probably becasue we suck at using it..
elson: Would one of your amps be usable as a DC spindle drive 107VDC @ 16A peak to use a treadmill motor.
anonimasu: no I didn't see fenn's stuff, is it online somewhere?
with 100x100 cubes it kills the computer..
'raw' OpenGL might be faster than vtk?
[20:09:34] <alex_joni> http://fenn.dyndns.org/pub/irc/matchstick.pn
[20:09:48] <alex_joni> http://fenn.dyndns.org/pub/irc/matchstick.png
how are the matchsticks rendered? 6-sided cuboids?
unless vtk dosent render non visible polys..
awallin-VM: still it shouldnt be that damn slow :)
03alex_joni 07v2_1_branch * 10emc2/configs/hexapod-sim/ (README core_sim_6.hal): loadrt genhexkins needed now
03jmkasunich 07TRUNK * 10emc2/src/hal/user_comps/vcp/vcp_main.c: add a 'use pyvcp instead' notice to halvcp
03jmkasunich 07v2_1_branch * 10emc2/src/hal/user_comps/vcp/vcp_main.c: backport: add a 'use pyvcp instead' notice to halvcp
03jepler 07v2_1_branch * 10emc2/configs/sim/ (lathe.xml lathe_postgui.hal lathe.ini): this finishes the backport of pyvcp (I hope)
03jepler 07v2_1_branch * 10emc2/debian/emc2.files.in: this finishes the backport of pyvcp (I hope)
03jepler 07v2_1_branch * 10emc2/src/Makefile: this finishes the backport of pyvcp (I hope)
03jepler 07v2_1_branch * 10emc2/src/hal/halmodule.cc: this finishes the backport of pyvcp (I hope)
03jepler 07v2_1_branch * 10emc2/src/hal/user_comps/Submakefile: this finishes the backport of pyvcp (I hope)
03jmkasunich 07TRUNK * 10emc2/docs/man/man9/pwmgen.9: add manpage for pwmgen
03jmkasunich 07v2_1_branch * 10emc2/docs/man/man9/pwmgen.9: backport: man page for pwmgen
03awallin 07TRUNK * 10emc2/lib/python/pyvcp_widgets.py:
- add dial widget by tomp. I renamed a lot of the variables to make
the code more readable. I'm not sure the key-bindings will be intuitive
for first time users...
- add timed update for jogwheel (fixes problem where signal connected to
pin was not updated before an event ocurred)
XML for dial is now:
tomp_: SWPadnos: if you want to add to pyvcp, please do a cvs update to get the latest pyvcp_widgets !
03jepler 07v2_1_branch * 10emc2/debian/emc2.files.in: package newly added files
23:12 < ds3> morning
er.. not yet
you on the other hemisphere?
is that around germany/denmark?
a bit more east
it's like "dark and stuff" here
where are you, eholmgren?
that may be due to the clouds that are dumping snow on us though
just putting in overtime hours at work on the weekend
empty parking lot was fun for about 15 mins before I actually came in ;)
in a large minivan no less
doesn't ABS/Traction control prevent that?
it's electronic 4-wheel drive
normally rearwheel only
unless it detects them spinning
but being more on topic, any idea what I am missing with the lathe simulation - it dies on the import vcpparse?
oh those.... the subarus are like that
ds3: trunk or 2.1 branch?
packaged, installed, or run in place?
one sec... this is run in place sim only
trying to cvs stat something
dumping snow in colorado currently too
(cvs status works better on files that are checked in ;))
hmm, that's been working, let me try
I just did a cvs up last night, should I do another one?
03alex_joni 07v2_1_branch * 10emc2/configs/hexapod-sim/README: instruct how to specify genhexkins
ds3: works for me. be sure you cvs update -dP in the toplevel directory (NOT src)
what's -dP? delete old files?
I did it on top at the check out point
cvs -H update says what -d and -P are
there is another one - import xml.dom.ext
is that third party?
-d is create new dirs, P prune empty dirs
03cradek 07v2_1_branch * 10emc2/tcl/bin/pickconfig.tcl: having a writable, personal nc_files directory helps a beginning user.
ds3: on ubuntu, installing the package python2.4-xml should make 'import xml.dom.ext' work
jepler: doing a fresh compile of everything. had to download and install PyXML. but that solve it.
heh... HAAS lathe programs are definitely not happy hehe
03jepler 07v2_1_branch * 10emc2/debian/control.in: pyvcp uses xml.dom.ext, provided by python2.4-xml
03cradek 07v2_1_branch * 10emc2/.cvsignore: some new ignores
03cradek 07v2_1_branch * 10emc2/bin/.cvsignore: some new ignores
03cradek 07v2_1_branch * 10emc2/configs/sim/.cvsignore: some new ignores
03cradek 07v2_1_branch * 10emc2/docs/src/hal/.cvsignore: some new ignores
03alex_joni 07v2_1_branch * 10emc2/configs/sim/.cvsignore: ignore scope cfg file
is the feed in lathe mode still IPM?
feed is in whatever units you like, depending on whether you program G20 or G21
inches or mm
not metric vs english; IPM or IPR
most lathecontrols default to IPR unless commanded otherwise
I understand the question now
that explains why the similation is taking forever ;)
feeds are per time (so far) in emc
the spindle speed thingie on the sim is a nice touch!
ds3: it's part of a thing called pyvcp (python virtual control panel)
users can extend that with all kinds of nice things (buttons/dials/leds etc)
alex_joni: it looks cool
is it correct that I need to do a G43 Hxx in the current lathe implementation to pickup the Z offset?
yes AND you must also be in G18 (XZ) plane
um, I think
to use shape compensation you do, not sure about tool offsets
tool offset is a bit more important for it to be useful; shape compensation can be avoided as long as all cuts are at prependicular
well we have both
you mean they are both implemented?
it's hard to cut an arc (the one you want at least) without tool shape comp
ah, the page says otherwise
did you see that you can define the tool shape in the tool table
for instance in sim/lathe load tool 2 (t2m6) and you can see it in the gui too
is there a GUI for the tool table?
no, was looking for a gui to edit it
after you edit it, you can reload it with Machine/Reload Tool Table
where is the text file for it?
whoa this more detailed then on the haas
[22:07:27] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?Lathe_Advanced_Features#Lathe_specific_radius_compensation_done
this picture shows the available tool orientations
DOH. the catch there is '(done)' where as on top it says "not yet implemented" in bold :/
this is a lot of stuff to learn for one session. enough for the moment.
* Jymmmm decided to say screw it, whipped everything out and is organizing his nuts
* Jymmmm thought bout putting some nails in his nuts, but decided againest it.
* Jymmmm put the nails in the ballcock instead.
Gee, I want vista now... :http://slashdot.org/articles/07/01/21/0042202.shtml
K`zan: hope you're beeing sarcastic
(for your own sake)
War is Peace. Freedom is Slavery. Ignorance is strength.
hrmmnn... maybe the surgeon shouldn't be watching movies while they're operating?
but they have to watch the instructional videos ...
the medical scenario is a red herring
SWPadnos: Oh come one, gimme a break... Heart Transplant for Dummies.... Learn Cardiac Bypass in 24 hours.
the point of the original article is how totally ridiculous the entire scheme is and how it will never work
why do you say that?
(the red herring bit)
more accurately, MS
GAWD lets hope they aren't using M$ in the OR
uh.. because people visualize MRI data on special purpose workstations?
not entirely true
MS's claim that the medical example isn't true is a red herring
and it's not just MRI
i'm saying it's a small market segment and isnt worth arguing about
exactly. I was talking to a friend who works for GE Healthcare, and when I mentioned the original article to him, he got very distressed.
messing up an entire general purpose computer in order to watch DVD's is ridiculous
he actually ended up writing a letter to their MS rep (they are responsible for probably 100000 licenses/year), asking what the deal is
alex_joni: Totally sarcastic...
messing up an entire general purpise computer in order to make sure people don't watch DVDs they shouldn't is even worse
jmkasunich: that's what i meant, sorry if my english is poort
the "Physician's Desk Reference" is content-protected media ...
(if it has videos)
I rip DVD's. Give the ripped copy to the nephews and reburn another one when they scratch it.
it's not just surgeons watching "The Abyss" while operating that's the problem
Toss the original where they can't get at it
As flakey as m$ stuff is I can't imagine anyone insane enough to run anything mission critcal on it...
and there are vids of blueray cracks allready posted on youtube
K'zan - but they do, all the time
unfortunately, there are many places where Windows *is* used, even though it's a bad idea
Fair use isn't anymore, need more $$$ to let you even near it.
i didnt even know there were any blu-ray discs out yet
warships, nuclear plants, high energy physics labs, hospitals, governments ...
Well, jmkasunich there is a lot of rampant insanity floating around these days...
* eholmgren cracks out
yeah - the first HD-DVD, "Serenity"
to show the hardware
* fenn puts on his tinfoil cowboy hat
Last warship I heard of trying that had to be towed back to port<shudder>...
good movie - I'd like to see it in HD
K`zan, yeah that's the one :)
Frightening. The very idea of m$ in my car scares me shitless, esp with the loony drivers in scabattle...
* K`zan wonders why people think it strange I still have my 84 caddy :-).
lets all revert to linux micro kernals on PIII's
works for me
before M$ got in too deep
Too depressing to think about this early in the day :).
THis, however, might be cool: https://wiki.ubuntu.com/UbuntuStudio
couldn't find anything of interest on ePay
[22:34:54] <alex_joni> http://windowsvistablog.com/blogs/windowsvista/archive/2007/01/20/windows-vista-content-protection-twenty-questions-and-answers.aspx
not one positive message on that page :)
I mean the user messages
how unexpected ;)
Comes as no surprise to me, m$ and the entertainment mafia are getting more and more into the fortress mentality as time goes on.
Bad part is that both can call down the men with guns...
Waking up and smelling the shit burning seems to be completely beyond either :-(.
You know the us.gov did the same thing with GPS signals... distorted for the masses while pure signal was only offerred to approved parties.
Skullworks: That I could see reasons for at the time.
and eventually that was abondoned.
Yep, when the need was 'no longer there" heh...
the error was not enough to matter
But in a real pinch they still can.
Well, who knows what they can do that the mil receivers can handle?!?
Skullworks: the error was up to a couple hundred meters
thats enough to matter if you are guiding a non-nuclear weapon
I guy I went to HS with is now an EE in the GPS industry
Would like that lot if the bad guys were using it to plot fire on my position :).
Not entirely true. The dropped the dithering, but the very high precision stuff is still encrypted. I believe it is 10 updates a second for us, but 100 updates per second for the military. (Or is it one update and ten ?)
"Sorry, I won't allow any content producer to take over my hardware even if this means installing linux as a second OS just for media (never used linux before actualy so I'm guessing it might do the job...)"
how exactly do they prevent civilians from using the military signals?
lerman: one and ten
fenn: encryption I think
I am less enthralled with m$ as time goes on, I'll keep xp just for the flightsim stuff, but that is it anymore.
i thought it was all based on phase shift.. ehh. nevermind
alex_joni: actuallu you will have no choice..
It would be great if a mbd mfg would make a run of linux optimised main boards... ( I can dream still)
jmkasunich: Caps arrived yesterday, THANKS SO MUCH!!!!!!!!!!!!!
soon traces and stuff have to be certfified..
to be m$ certified..
Nannies neverending, sigh....
I heard that drivers must be m$ certified to run under vista (making an already bad situation worse)...
How do you get them certified if you can't install and test them?!?
I guess you just pay $$$ for the blessings....
anonimasu: have no choice with what?
""No. HFS uses additional chip characteristics other than those needed to write a driver. HFS requirements should not prevent the disclosure of all the information needed to write drivers."
worse - no more unified drivers - each unique device must have its own approved driver - hope the mfg does not go belly up or drop support for your hardware.
I am interesting in what problems might arise when a programmer reverse engineers the non-disclosed information about the hardware. Does the programmer get into legal trouble for using the undocumented features on a piece of hardware that he has paid for?"
Just bend over and pay, pay, pay....
they apply the DMCA
Dunno what the truth of it is, but I have heard the business community is down on vista big time.
DMCA is protectionist evil of the highest degree...
to reverse engineered drivers
It's simple.... don't use Vista.
I have no intention of doing so.
The only thing linux doesn't do that I like to do is flightsim and that is a fricking GAME :-).
FGFS and X-Plane, not quite the same, but ....
K`zan There's always XP for that (a la wine?)
I run W2K on my Dell 530 dual 2.8 xeon system for Solidworks and my CAD/CAM.
sadly, most high-end (but not super high-end) CAD software doesn't run on Linux
Jymmmm: Nothing I ever wanted to run, runs under wine :-(.
Couple editors is about it.
Someday I hope that there will be a suitable linux alternative
And that varies from release to release :-(.
and wine has a problem on 64-bit machines (though I should try again, I think there have been recent improvements there)
K'zan: you're welcome (about the caps)
Jymmmm: Way too expensive and requires the original anyway.
vmware is free, for some versions
Skullworks, with an openGL card?
SWPadnos: Far too many things either don't run or aren't available under 64 bit anything at this point - I've reverted to 32 bit stuff on the 64 bit box here.
and if you have an old Win2K, you can install that (though that's probably against the license, even if you uninstall it from the original machine)
but VMware takes such a performance hit
it's not all htat bad, for some things
can't wait for Gutman's response
For something like flightsim, you need all the horsepower you can get as it is...
I have an installation of SolidWorks on Win2k in VMWare, and I can spin medium to large models in realtime
Skullworks, and if so, does the openGL card improve solidworks any?
err - SW2004
OpenGL is the best
which card do you use?
yeah, and VMWare has a good openGL driver
I also have a pair of FX3500s I intend to rty
FX3500s I know
NVidia 7800GT, 256M, manufactured by XFX
I run an old Quadro 900XGA
not the 7800GT
I was going to go for a Quadro
but it's very fast, like 13800 FPS in glxgears
of some flavour
the quadro was actually singificant;y slower
Quadros are not speed demos for frame rates
stability is important to me
but the shading...
like 1/3 the speed (though I never changed X drivers)
SW2006 occasionally craps out on me
I always blame the shitty onboard Intel 810 grpahics
I have 2007 - but have not installed it on the "BETA" box
wait a couple of years - it'll require Vista ...
they moved a bunch of stuff around and changed icons... so there is a real learning curve on this upgrade
not looked at that yet, im out of service contract at the moment on SW
is sheetmetal any better on 2007?
I was too
was still on 2004
but had a few die jobs that made it worth it
see you LH
had my die engineer use it to do the job - but have not moved it back.
[22:59:45] <alex_joni> http://www.gamasutra.com/php-bin/news_index.php?story=12314
I was hoping that Blender might someday mature into a useable ware for making geometry that could then be posted as G-code... but I think its kinda died off.
alex_joni, so, here was my plan
Wish there was something for that, there is qcad, but I still can't imagine what one does with 2D...
20mm steel plate, 500mm x 700mm ... welded onto some 100x50x10 C channel
I was thinking of welding it directly to the channel
I could make a raised base from some 10mm plate, vertically, so spread the load over some more of the C channel
the 10mm plates would have to go in X and Y to keep it stable I guess
what's the channel for?
the cell base
so not just one channel..
directly under the bot say 2 channels runing full width
then 2 running front/back
sounds good to me
plus a few more to make up the floor, with aluminium checker plate as tread plate
what thickness is the channel's width?
its 100mm wide
the walls are 10mm thick at the sides
oh, sturdy enough
I think 7mm on the 100mm back face
its either that or I section beam
I think C is better than I beam
I guess what is crucial is the stiffnes between the 'bot mount and the work table
don't think so
distance & co is more crucial
table won't move around I hope :)
wow, is /. always so clueless? "
wow, is /. always so clueless? "IT: Chinese Prof Cracks SHA-1 Data Encryption Scheme
"she and her team have taken out the SHA-1 scheme, which includes the (highly thought of) MD5 algorithm"
jepler: it's filled with people that want to be geeks..
Skullworks: blender hasn't died off at all, it just was never suited for cad modeling in the first place
blender's main strength is its interface, which is crap for anything precise
alex_joni, my theory was to make the coupling between the table and the bot ultra rigid, so even if the bot managed to flex the floor beams, the table would move with it.. thus maintianing the relationship between table and bot
are these not in the vtk distro package?
fenn: I dont know
fenn: maybe somewhere
fenn: it looks like they make actors for all objects
but there's only a few parts
robin_sz: you might be overengineering it
oh that spline is sexy
alex_joni, woah .. you are the one that uses 40mm plates not me :)
but yeah, its only a baby bot
those are 3t turn/tilt tables
this is only a 350kg bot ... with an 8kg payload .. nothing really
just bolt it to the concrete, sheesh
that is so not an option
we do do something wrong!
what did you have in mind?
wait a sec..
robin_sz: guess your plan is way OK
I'll let you know what it looks liek when we get it built this week
anonimasu: i'm gonna try stuffing a vtk array like i said and see how it works out
wait a sec..
alex_joni, you recon on solid sides or 25mm mesh for the cell?
* fenn waits
25mm mesh + protection you mean?
[23:29:41] <anonimasu> http://public.kitware.com/cgi-bin/viewcvs.cgi/Examples/VisualizationAlgorithms/Python/ExtractGeometry.py?rev=1.3&view=markup
whatever is cheaper :)
usually (over here) solid (thin) sides are cheaper
I used sample.SetSampleDimensions(200, 200, 50)
that's still pretty slow
that's a crapload of cubes though
just on generating..
but it's a crapload :)
heading for bed here
good night all
spinning around and stuff is smooth
alex_joni: no :/
I need to save it and copy it to the work server
im half into bed :)
well.. it looks like .. something
and I like the orange
that's just an example that ships with vtk
I would have modeled a sheep as an example
they really do add actors for each object.
[23:35:13] <anonimasu> http://public.kitware.com/cgi-bin/viewcvs.cgi/Examples/VisualizationAlgorithms/Python/officeTube.py?rev=1.2&view=markup
what are you saying with this last one?
look at how they add actors for all objects..
but there are only like 6 objects
in the gazillions of cubes see where it says dataMapper = vtk.vtkDataSetMapper()
dataActor = vtk.vtkActor()
there's only one actor for all those cubes
maybe mailing the userlist would be a good place to start
i dont think its that hard
i would like to find some better docs than just doxygen
that's the problem really
looked at this? http://www.imaging.robarts.ca/~dgobbi/vtk/
its a start at least
yes that's where I got my cone code :)
I need to go to bed now