ok, can do that
jepler, thanks, that will keep me busy for a while. though i think a faster cpu is due.
hmm I am seeing the wierdest thing with ddts ...
I have a ddt which I feed the -fb of an axis into and 90% of the time it gives a reading of 0 but just for a split second with 10 sec intervals it goes to what looks to be the correct value
Rugludallur is now known as Dallur
Anyways, I will solve it tomorrow, good night everyone
03cradek 07HEAD * 10emc2/src/emc/rs274ngc/interp_convert.cc: stupid bug
YO-YO's are for kids.. like trix...
03cradek 07HEAD * 10emc2/nc_files/useful-subroutines.ngc: fix incorrect entry arc
03cradek 07HEAD * 10emc2/src/emc/rs274ngc/interp_convert.cc: same bug, second try
* fenn brutally slays gcc-3.4.3!
I'm about to burn the coolcncb05.iso in order to check out your software and try to contribute in code and hardware development
I'm an electronics engeneering student, the last thing I did was a PCL812 data acquisition card software that plots voltage like an Osciloscope
chinamill is now known as Eugenics
Eugenics is now known as eugenics
03cradek 07HEAD * 10emc2/configs/sim/axis.ini: allow loading python scripts that generate gcode on stdout
cradek: tell us about "allow loading python scripts..."
ok, make a python script that outputs gcode to stdout
in AXIS, hit open, select your python script
cradek: no, it's not quite that simple
yes it is, I just did it
did you miss an obvious way to use filters?
+PROGRAM_EXTENSION = .py Python Script
+py = python
but that's not in any of the standard .ini files
look up ^^
oh I missed it
* jepler hides
do you think it shouldn't be there?
seems very useful to me
yes, if you only click .py files that you trust
so by clicking the "go" (for instance) button on the python script, the g-code goes directly to the machine and runs?
when you load the file, it runs the script for you and loads the output into emc
after that, everything works normally
if you change the script, you can hit reload in AXIS to load the new gcode program
loads it just like 'open' in auto mode
yes it's just a special kind of 'open'
in AXIS you can specify filters that process the file into gcode while opening it
for instance you can set up a filter for images (gif,png) or dxf or whatever.
slippery slippery slope
because of 'haas' ?
no, just because pretty soon you'll be implementing toolpath algorithms
sure, image-to-gcode.py has algorithms
anyone can easily write filters
hmm.. i guess what i'm saying is that since this is starting to look like a cam system maybe we should figure out a good shared library/api
for stuff like selecting which tools are available
so far the API is: passed a file on the commandline, outputs gcode on stdout
i suppose axis could get that from the tool table but its lacking a lot of information
axis does NOT generate toolpaths
axis loads gcode that can come from a variety of places
as a separate issue though, I wonder if we should add more info to the tool table like I did for lathes
the tool table should probably be completely overhauled and put into XML with the rest of the config files
cradek: Hurco sued HAAS a few years ago for including 'conversational programming' with their controls without paying royalties
I just wasn't sure how that might apply to open source
maybe if the hooks are there, but the scripts are user generated, it might be ok
that might keep you from having to go down the 'slippery slope' in supporting users on g-code generation
dan_falck: i think that whole thing is bullshit
fenn what whole thing?
yeah it sucks
hurco suing haas because of 'conversational programming' in their controls
Eh, then file a patent/trademark in a timely manner =)
you shouldnt be able to patent something so obvious
patents really suck nowdays
Same thing with Unisys and the GIF compression algo.
its like, patenting the use of CNC machinery for "making stuff"
fenn but if you designed some new tool and have invested $40,000 you'ld want some ROI
pff your ROI should come from PAYING CUSTOMERS
go the extra mile and market it successfully
not from suing other companies
and not just have some shumck steal your idea right out from under you.
you can't steal ideas
ideas are worthless in and of themselves
fenn your new tool.
problem is, there's a lot of people who have the same ideas at the same time in different parts of the world
gee, wuoldn't it be neat if we could have a machine that just did whatever i wanted
that's basically what "conversational programming" means
obvious leaps in thinking from looking at other ideas
They could have came up with a different name
and made it JUST different enough
now, if haas reverse engineered hurco's algorithms there would be a problem
still, i'm all for standardization
fuck greedy patent hogs
whatever happened to capitalism?
fenn: ok, so you wouldn't mind me getting your new tool mass produced in china then outsell you by 75%
that's why I think coming up with the invention is only half the job done
Jymmm: if it were a free market
fenn come on now.... one side of the fence or the other here.
what? the patent system is protectionist.. i'm advocating free markets
R&D is VERY expensive.
and a crapshoot to boot.
lawyers+large amounts of cash= patent
that's why nobody does r&d anymore :P
fenn Oh so that's what LEs is NOT doing...
20years isn't so bad, after that everyone can benefit
20 years of the most obvious thing is ridiculous
and most patents these days are the most obivous things
Look at the guy that patent online auctioning, ebay had to pay them. He was smart enough to file.
iirc is the payment system, dont recall.
and you think that's a good thing?
he was dumb that he didn't get 35% shar eof ebay.
how about drug companies patenting ancient Indian herbal medicines
how bout drug companies patenting the genomes and proteins of animals and plants, including humans?
15 years of drug R&D and they could come up with shit. that gets expensive.
yes, someone has a patent on _your_ hemoglobin
I will say it's a bit pathetic these days, but protecting your IP is a good thing.
fenn: on a different subject... how's the CAD/CAM hunt going for you?
dan_falck: i've been wallowing in STEP documents and trying to learn about XMl
XML is evil!
Jymmm: dont even start
fenn too slow biotch!
i'm hoping to work with brlcad to finish a portable step import library
fenn:I have been studying scripting (python) and how to interface it to C functions
I'm taking small steps
dan_falck: i'd definitely like to make a python interface to the step library
such a thing doesn't exist anywhere yet
* Jymmm files a pantent
03cradek 07HEAD * 10emc2/nc_files/useful-subroutines.ngc: slot cutting: call me crazy if you like
* fenn patents slot cutting
prior art ^^^
fenn: I am just messing around with some of Jon Elson's C routines
and looking at swig
* fenn patents the idea of a prior art database
* Jymmm copyrights the nick name(s) fenn and all if it's dirivatives.
good luck against fenn mfg. co
I patented the letter 'J'
fenn fenn mfg has nothing to do with internet communications.
EMC has nothing to do with hard drives
remember copyrights have catagories (well, reg tm do)
its still some bull crap you have to worry about in case a judge gets a wild hair up his ass
(and some money in the pocket)
* Jymmm lets fenn vent in peace
dan_falck: have you connected jon elson's c programs with python scripts?
I think I can though
that would be neat as a proof of concept
fenn:OpenCascade has a STEP library in it
although with psyco i dont know how much of an advantage writing stuff in C would be
I just read that this morning
dan_falck: opencascade isn't free, it has a wacky license
and its really hard to understand
yeah that's true
supposedly opencascade was derived from a work by nist, which is still in the public domain
what did I miss?
but i think it'd be better to latch on to an existing open source project, like one of the 3d modeling packages
k3d has the foundation for a really slick parametric procedural modeling system
sean said they had the ISO standards documents for the most useful step AP's
and are allowed to share for work directly related to brlcad
terry ridder actually started a step import library and made some decent progress
er, brlcad has the iso standards documents i mean
cradek: you're crazy
jepler: I know
I almost put the .py in nc_files but it seems like that's just asking for a fight
maybe you should make an interface for calling .py files from gcode
or embedded .py scriptlets
* fenn points at jepler
you can embed gcode in python
using a special command called "print"
you can also embed a text editor in axis..
fenn: it's clear to me that you have a vision exceeding my own. I can't wait to see the results of your upcoming work. It's sad to see a fork of AXIS, though.
i'd call it "Joint"
as in "I could use a "?
* Jymmm patents the use of "echo" and/or "print" as used in a programming language =)
jepler: see ya
jepler: It's ok to talk to yourself, but let us know when you start answering yourself too!
dan_falck: psyco is a python module that "compiles to c" or at least makes it run a lot faster the second time
I'll check it out
03jepler 07HEAD * 10axis/debian/control:
optionally use psyco, which seemed to speed up loading a large file by about
25% (6 seconds instead of 8 seconds).
03jepler 07HEAD * 10axis/scripts/axis.py:
optionally use psyco, which seemed to speed up loading a large file by about
25% (6 seconds instead of 8 seconds).
fenn: thanks for reminding me about that
psyco has a lot of cool stuff in it
it's very much like how i imagine AI systems should work
it has to decide what to compile and how to optimize it without optimizing out all the important details
this explains it if you're interested in the details: http://psyco.sourceforge.net/theory_psyco.pdf
asdfqwega: what documentation about axis on bdi4 were you having trouble finding?
The link you sent on ourproject.org
[19:30:54] <asdfqwega> http://ourproject.org/moin/Rebuilding_python-axis_package
asdfqwega: that page loads for me
jepler: It comes up with a "Configuration Error" page for me
Make a liar out of me...it just came up
I wonder why
jepler: argh, I'm getting joint following errors
commanded Z velocity is twice what my ini says
and it's not even a reversal, it's an X-only move into a Z-only move
oh no it isn't, there are two moves, it's colinear in Z
[20:15:29] <cradek> http://pastebin.dyndns.org/307
line 4 is the problem move, see it's Z only and maxvel in Z is .3667
cradek: uh oh .. does reverting my "don't always halve acceleration" change fix it?
but it's a vel problem, not accel
that's why I'm puzzled
jepler: Is there any limitation to the highest version of AXIS that would work with Paul's emc-1.49?
I've just tried working with AXIS 1.4a0 from CVS
asdfqwega: I haven't regularly tested for compatability, for obvious reasons
asdfqwega: I had hoped that the latest CVS versions would work, but really have no idea
jepler: all those maxvels look suspicious
which field is maxvel?
the last 3 are 100% vel, max FO vel, accel
line 3 is a move in X only, although you can't quite see that for sure since the previous move was an arc
you can see its max FO vel is also way too high
hm, I notice those fields aren't shown for the circular move. are they in the message?
yes if I'm looking in the right place
cradek: can you send me the file?
can I run this on a sample config and get the error?
[20:29:38] <cradek> http://timeguy.com/cradek-files/emc/lathe2-experiment.ngc
to be run on .. nist-lathe?
this file uses radius comp but if I disable it by loading t0, the same thing happens
no, try stepper-inch
for max you'll have to set a Z offset (jog up a bit, shift-home)
I'm running on stepper_inch.ini
is the error during the "slot"?
it's the next to last move (g0z safety)
yes bug happens on max
it didn't on stepper_inch.ini
maybe different axis vels are necessary?
arrgh, it jumps when I turn machine on after the FE
I thought that was fixed
yes, I get it too on max
man.. reading about the haas/hurco conversational programming patent lawsuit just makes me mad
and in other news, walmart is attempting to trademark a smiley face
yeah I nearly had it distilled too
does not FE on 2.0.1
are the messages sent the same in accel, vel?
in 201 they are just what you'd expect, vel .3667, accel 20
Issuing EMC_TRAJ_LINEAR_MOVE -- (+220,+88, +0,0.000000,0.000000,0.193300,0.000000,0.000000,0.000000, +1,0.366700,0.366700,20.000000,)
Issuing EMC_TRAJ_LINEAR_MOVE -- (+220,+88, +0,1.000000,0.000000,0.193300,0.000000,0.000000,0.000000, +2,0.374851,0.374851,20.444586,)
the vel on that second line is wrong too
so is accel
that's the g1x1 move in fe.ngc
reverging the "merge segments" code fixes it
cvs update -j1.60 -j1.59 emccanon.cc
and then manually add an empty FINISH() which is used by some other code
drat? it means you're off the hook!
I was thinking it was the message-combining
yeah but it means it's a harder bug than I thought
getStraightAcceleration x=1 y=0 z=0.2222
getStraightAcceleration x=25.4 y=0 z=10.2159
anything look fishy to you?
umm let me think
03jepler 07HEAD * 10emc2/src/emc/task/emccanon.cc: canonUpdateEndPoint must be called with internal length units
cradek: I think this fixes it
though now I'm worried there's a bug if you switch units during the middle of some chained segments
fe.ngc and lathe2-experiment.ngc both run now
go Italy go !!
we are the champions my friend :P
love this song
sorry for geman friends, we played a better match
yes, looks very promising
thanks for fixing it
I'll use it to make a part after dinner
cradek: what about your duty to set off the "fireworks"?
you mean eat meat and buy chinese explosives?
everyone needs a hobby, but mine's not that
oops, .1875, not .1835
3/16 is .1875
I still get those wrong sometimes
so write [3/16]
could someone help me on the BDI installation? I seems to get telnet login screen when I restart after installation.
unfortunately I don't know much about BDI these days, I use ubuntu
I do not understand linux :(
I tried ubuntu but because my machine do not have access to the net, I cannot install either
no internet? bummer
I wonder if you could direct me to where and what package to download to install the RT
you could take the emc2 packages to your machine, but easier might be to take your machine to the internet
I am on a net cafe now
what RT? BDI already has a RT kernel
ah I see
yes BDI has RT but as soon as I reboot after the installation, I get into a telnet login screen and nothing happened.
I could login but I do not know what to do next
you should probably ask on the BDI mailing list
(unfortunately I don't have that address handy)
the packages for emc2 on ubuntu breezy are at http://dsplabs.cs.upt.ro/emc2/dists/breezy/emc2/binary-i386/
of course you don't need all the versions, just the latest of each package
cradek: thanks I will try
if you're not sure which version of a package you need, ask and I will help
ok which RTai should I use?
3.3-1, 3.3-2, 3.3 -test4 ?
do you understand these are for ubuntu breezy, not bdi
you _really_ don't want to do your own install of RTAI
I have no problem with reinstalling ubuntu
this is separate from bdi, a different alternative for running emc2
ok I just wanted to make sure you knew this was not for bdi
I will reformat the BDI installation and do ubuntu instead if it could help me to start
once the ubuntu install is done I believe you could copy the needed packages to a CD and take them to the PC
in my opinion that's a good idea if you want to run emc2
"needed packages" includes packages for RTAI and emc2
you will have a very hard time getting bdi to build emc2 without internet
jmkasunich: we're a step ahead of you, he's looking at the ubuntu emc2 repository
okay, BDI is out and ubuntu is in :)
ZZhung: you want rtai-modules-2.6.12-magma_3.3-2+cjr12_i386.deb
ZZhung: also linux-image-2.6.12-magma_cjr12_i386.deb
I am tagging them for download now
WHat else do I need?
and you need bwidget
if this list is out-of-date, it should be fixed: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#Installing_and_upgrading_off_line
hey look at that, documentation
(I've never used those instructions, they were prepared by someone else)
okay at the wiki page huh?
I'm updating that page
where do I get the lib packages?
from the ubuntu pool, just a minute and I'll find a url
I'm not sure why libc6-dev is on that list
it's unlikely that you need it just to run emc2
(-dev indicates it is for software development)
okay, but what about the other lib packages? there are about 6 of them
you definitely want the libgtk
usually, you need libXXX to run a program, libXXX-dev only if you intend to compile it
k.. I am now googling the lib files
[23:51:42] <cradek> http://us.archive.ubuntu.com/ubuntu/pool/main/g/gtk+1.2/
cradek: tha :)
I cannot find