Jim Triggs is coming out this way in February
we should set up a CNC router for him sometime
You making parts for him now?
I pulled my old gantry router out of the corner last night
time to set up another emc box
I'm playing with networking right now
I'm installing 6.06 on a mini itx right now
ssh seems to work okay for me
I set up an older computer in the garage with emc2 and have been ssh'ing into it
from this desktop
right now I'm playing with qemu a bit
still have to use Autocad sometimes
wine is getting better for that though
That's good to hear.
I've not tried wine at all.
banished the last of my windows machines a while ago
wine is able to install autocad on a ext3 file system?
rhino and vectorcam too
Are you able to use all of vector?
As far as I can tell, everything seems to work with vector now
you had problems with that a few years ago.
seems good now
I'm working on the desktop cam stuff a lot this week
I have a 'Shuttle' pentium 4 box here, in the house
pretty fast machine
connected to an older 1.0 Gig Amd in the garage
this machine (shuttle) really rocks
3.0 gig processor
I am setting up ubuntu 6.0... on all the machines here
with emc2 on the older slower boxes
That's about what I've done here.
by the way, I know how to get the 'open in terminal' thing happening in Nautilus now
had to install from synaptic
qemu is fun
fpga's are fun! so i hear.
what ever happened with that opencascade project?
we put it on hold for now
we are demoing UGS NX for linux/OS X at the moment
but, it's still conceivable that we might use some opencascade code
too many irons in the fire, as usual
DanielFalck: What was the package name that you added to nautilus to get the "0pen a terminal" ability?
I think it was 'nautilus-open-in-terminal'
I'll look no
once you're in the directory, right click
Hi roltek. Just got that mini-itx board running with ubuntu.
did you just pull it out of box
Yep. I've got one of those automotive cases.
what you use for a case for it ray
just used the back panel to mount it and a mini drive
the blue 1
Yep the blue one.
That case is made for the automotive power supply.
did you put a messa card in it
I used the tiny 12 volt direct plug in model
do you think it will fit
I believe that I could add a mesa with a riser card and the mesa folded over the mobo
But I'll have to try that later.
i was looking at them on the dsl site
Right. That's the one.
Nope. The two slot riser board is taller than the inside space in the blue box cover.
The top slot on the 2 slot riser will let the mesa fold over the mobo.
makes a really nice compact system that way.
chek your e-mail ray
That is all of it but the 12 volt dc supply.
aip_tom is now known as aip_tom1
aip_tom is now known as aip_tom1
i have a question for you
in this schematic, http://www.kg-uitdaging.nl/COlabvoeding/CO%20labvoeding%20V2%20schema.pdf
i'd like to know if the bottom rail in the picture is supposed to be below or above the ground level
i don't know
If, by bottom rail, you mean the lowest horizontal line, it is below ground level.
You can tell that because the positive side of C4 is connected to ground and the negative side is connected to that line.
I'm testing the new m5i20 configs Peter W sent me...
so anyone know how I can replace the fpga config with a new one ??
I now got the old one working OK
I'm trying to test some new m5i20 configs peter wallace sent me
but I haven't gotten very far
Ah. I saw that exchange.
Jon E's amps want 50 kHz, so 50kHz it shall be
Let me see if I understand your setup.
do you know exactly what happens when I say 'loadrt hal_m5i20' ?
mesa card producing pwm
I need to change the .bit file that is loaded to the fpga
pwm sent to pico amps
yes, pwm from mesa card
Back in the old days, before PeteV produced the stock mesa install
We had to load the fpga code using a little linux script that mesa wrote up for me.
I really don't know how to change the current fpga load and mesa run.
but is it done by the hal_m5i20 driver now ?
there is a program "m5i20_cfg" in bin/
er, no _
but there's also the load_fpga switch when loading hal_m5i20 ?
This selection of the "desired" configuration of the anything io card is a topic that needs to be addressed.
etla: ah -- yes, that loads the firmware image specified at compile time, apparently in src/hal/drivers/m5i20_HM5-4E.h
jepler: yes, looks like that...
I'm not sure what process transforms the .bit file into that header file
I don't think it's automatic
the .bit file is binary, I get all kinds of funny characters with 'more xyz.bit'
yep me too
I see that there is an arg during startup that allows not to load the fpga.
halcmd loadrt hal_m5i20 [loadFpga=1|0] [dacRate=<rate>]
and loadFpga=1 presumably loads src/hal/drivers/m5i20_HM5-4E.h onto the fpga
but that .h file is not in the same format as the new .bit files I got from Peter W
there's a utility: src/hal/drivers/utils/m5i20cfg.c
the bitfile converter is supposed to be in there as well, but it doesn't look like it is
it's a pretty simple program to write: read in bytes as binary, output bytes as hex numbers separated by commas
I don't even have a /utils on my HEAD checkout ??
src/hal/utils - sorry
right you don't have that one either, or right now you found it?
see you Ray
I now found it ;)
is m5i20cfg.c compiled during the normal 'make' process ?
yes I think so
I have a bin/m5i20cfg program
I'm not sure - I'm looking at the source on the web from Windows ;)
it looks like m5i20cfg HOSTM54E.BIT produces that .h file
oh, ok. I thought it would actually load the bitfile onto the card
it can do that too
.. I think
that's what it looked like the initial comment said ...
it actually outputs to the screen a whole lot of hex numbers.
I think the card number is not really optional when using m5i20cfg to program the fpga
the I get 'Successfully programmed xxx bytes'
I wonder if it actually programs the card or just outputs the .h file to the screen
I believe that in that mode it outputs the .h to the screen
so I'll pipe it to a file
give it a card number to actually program: m5i20cfg HOSTM54E.BIT 0
it looks like there's an option (called PrintConfig in the source) for that
m5i20cfg hostmot5_4eh.bit 0
gives me an error 'Can't enable direct port access.
may have to "sudo" since it directly accesses the hardware
you probably need to run as root
yep, now it looks like it worked.
ok - if argc==2 PrintConfig = 1, so leaving off the card number will print the bytes to the screen
Let's try with the highwer pwm rate... wait a bit
now the Hz calculation when running the driver is probably way off
previosly when you set 32000 it would turn out pretty close to 32kHz
now when I put 50000 I get about 18kHz
I thought the PWMs were fixed frequency
config dependent, not parameter dependent
you give hal_m5i20 an argument when you load it
the config determines a master clock
33MHz in the old ones
now I have one with 100 MHz
now I need a spreadsheet to calculate a bit...
eyah - it's a 16-bit number, with the base clock as a constant
pCard16->pwmRate = dacRate * 65536 / 1000 * 1024 / 33000;
if you want 50000, I bet you need to ask for 50000/3, or 16666
but I have 100MHz instead of 33
right, so all other things being equal, you should get a PWM rate that's 3x what you ask for
ok, I'll try that
asking for 50000 was probably too much, as about 97kHz is max with a 100MHz clock
hmm... what's the default behaviour for loadfpga ? I may just have overwritten my new config...
default is 1 :(
Yes! now it works, I measure 50.25kHz on the scope
the process of loading different configs probably should be made a bit easier if people want to use them in the future
I guess ytou need 16583 then :)
well, loadfpga=1 makes sense, since a driver will likely not work with the wrong firmware
the driver would also need some work to support the 8-axis cofig which is available
yep - it could auto-detect what's there, and barf if the wrong ID is found (or any non-supported ID)
maybe I should check if I can still do cvs update ? anyone know where the scratch.txt is for testing ?
it's probably better to do that, and have loadfpga=0 by default
er... cvs commit I mean
I think you're on the list - you got the mesa email, right>
ok. that was generated from the list of people with commit access, plus a few others
so would you say it's OK to commit these to HEAD even though they are not tested and loading them is a bit tricky
good question. Peter sent me a copy as well, but I haven't wanted to commit without being able to test
I also have the update that fixes the index pin error, but again, I couldn't regenerate the bitfile from source, and I couldn't verify the changed behavior ...
the change from 33MHz to 100MHz seems to be working
other than all the time scales being off :)
what was the index pin error ?
I would prefer to see anything committed to CVS to be buildable from source with the xilinx tools
Peter W sent the sources for everything also I think
actually, I Think it was the secondary indexes
at least there are a lot of .vhd files
yes, the secondary encoders were all over the place
he sent the sources, but I'm not sure anyone has actually generated an identical bitfile
or I mean they are, in the current config
or a provably equivalent one
probably not in the 100MHz config you have
yes, Peter wrote that he fixed the secondary encoders a long time ago, so they are probably OK in all the new configs
I'll try committing them and see what happens :)
make sure you regenerate m5i20cfg.h (or whatever it is) from the HOSTMOT5-4E bitfile
m5i20.h, that is
oops - m5i20_HM5-4E.h :)
I think I won't mess with the current config, just add the new ones to cvs so they are available for all
I wonder if separate dirs are warranted?
probably a good idea
I have four new .bit files
but impossible to remove, so I'd make sure someone other than me agrees :)
huh? impossible to remove ?
directories from CVS
though I guess jepler or cradek can just delete them from the repository
CVS keeps the dirs around even if you delete all files and the dir, because you have to be able to check out a previous version
(which contains it)
ah, ok. the primary index inputs are on the B connector instead of A, and the secondary inputs are kind of interleaved instead of on separate 4-bit blocks (due to a set of typos)
hmmm - that would be a problem for replacing the "standard" 4-axis config, since anyone who has a machinewould need to rewire their index mask inputs (if used)
hmmm - I'll have to look at that a bit more, later.
so how does cvs commit work anyway ??
you add or edit your files, then "cvs commit", verify it's the list of files you expected to be added or modified, then enter a message explaining the purpose of your changes.
cvs [commit aborted]: could not open lock file `/cvs/emc2/,TODO,': Permission denied
is this a developer checkout or an anonymous checkout?
aip_tom1 is now known as aip_tom
ah, there is a difference ?
"Permission denied" messages typically indicate that you have an anonymous checkout
OK, I see, I probably need a developer checkout
anon(ymous) can check out files, only developers can check in new files
have you gotten a developer release before on the machine you're using right now?
er, a developer checkout
yes, I think so, I modified some docs earlier
now it asks for a password, should it do that ?
no, only use of SSH keys is permitted
does the username need to be the same as my unix username ?
I'm logged in as 'etla' on this machine
it has to be whatever username chris used when he set up the CVS server
but I think my emc developer username was 'awallin'
then you'd use :ext:firstname.lastname@example.org/cvs
that gives me the password prompt
you may have to wait until chris is around, he'll be able to help you
the cnc_toolkit group have spawned some 5 axis hardware (A , B&C) http://www.doughtydrive.com/
looks like the wiring for the spindle would be a challenge
jepler: what do you think about doing VCP type stuff with Python ?
etla: I think python might be a good language for it, and if we have competing GUIs then why not competing VCPs?
I don't need a VCP personally so I haven't really done any work on it
OK. I'm just a bit frustrated by the lack of widgets currently in VCP. I can fairly easily do better with Python and TkInter
But I don't have the skills to write a parser for a .vcp file
so anything I write would be pretty specialized for my machine only
I'm not sure whether compatability with .vcp files is important
there are not a whole ton of them out there yet
but maybe it can be cleaned up later if it's useful
if you come up with something flexible in python+tkinter I will be interested in integrating it with axis
sounds good. I'll see what I can come up with.
so that they can appear in the same window -- placing the vcp widgets on a new pane on the right side, for instance
Now that I have the vmware thing on my laptop I don't have to sit at the cnc mill when developing
linux or windows as the host OS?
it even runs the RT kernel, no need to do any special sim or non-RT tricks
I use ubuntu and --enable-simulator on my laptop
and have XP inside vmware for the rare occasions it's useful
actually while working on pluto-p it's been a necessity
what's pluto-p ?
jepler: you have xp running within ubuntu?
it's a parallel-port FPGA board.
about the board: http://www.fpga4fun.com/board_pluto-P.html
what I've been doing with it: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/docs/src/hal/pluto_servo.html?rev=HEAD;content-type=text%2Fplain
can you remind me how I generate the ssh key files that I need to send to chris ?
ssh-keygen -t rsa
then you send the ~/.ssh/id_rsa.pub file to chris
you can leave that blank if you like
can these keys be used on many machines ? or just the one their generated on ?
you can use the same identity on multiple machines by copying .ssh/id_rsa to the other machine
etla: pygtk and/or pyqt have the widgets you need? ( i made analog vscale for VCP, like a slide pot that ties to an analog output, maybe thats useful )
tomp: I have not looked at pygtk/qt. are they included in the default emc2-dev packages ?
no they are not
so it might be a good idea to use only tkinter, since it's there by default
they have lotsa widgets & interface to python, allowing interpreted consoles that connect to HAL. I've made some simple things work for the STG2.
if std toolbox is important, then just use tk tcl tkinter python & c ;)
it looks like python2.4-gtk2 is part of the base installation of ubuntu, so it's not an extra requirement for most users
but only python+tkinter will integrate with axis -- so consider that when deciding
jepler: i gotta study the new links you provided this morning to cadcamdro, like 'jdi', thanks
I have a 100 mm slide pot in my junk box....
.15 % linearity....made in Switzerland...
That is an acuracy of +/- .006 inches across the 100 mm ...
The m5i20 manual says that there is C code in the zip file that can be compiled under Linux.
in fact there is a /UTILS/LINUX/ directory with the code in there.
to test the board or realtime ?
To upload a bit file to the fpga.
In fact mesa compiled that into an executable in that directory.
utils for etla :)
rayh: I got that sorted out already. the utility comes with emc2
the new config works fine, I get 50kHz PWM
I saw that from your post.
I will answer that post with a couple of questions.
What should happen if I have more than one m5i20 aboard?
wb9mjn: are you sure that linearity == resolution
we probably need a new parameter <boardnr>
all the HAL pins have the stupid zero there anyway !
What procedure should be added to the loader to assign specific configs to specific boards?
Yes they do.
can you load hal_m5i20 only once ?
The reason I ask is that I've got a 2 board controller in process.
I'd think that if you tried it twice it would say it's already loaded.
for the mazak ?
The mazak would be an excellent example of a two board solution.
we used an older ISA card along with the motenc
the pwm rate could probably be made into a parameter ?
instead of being specified when loading the driver
Yep....just like an encoder, only linear....
Its a linear pot...has a rod that goes back and forth to operate the wiper contact....
wb9mjn: but I'm not sure that a good linearity means equally good resolution/hysteresis etc.
.006 is not great resolution...
So, I think yes...its probably that accurate...
Probably near the practical limit for linear pots....
If the wiper is connected on two ends, with the center contacting the resistive element, hysteresis would be
Connections on ends of wiper inline with motion...
so do you have any practical use for it ?
Nope....came out of something I scrapped, do not even remember what....
Probably a printer....
although 4 inches is too short for a printer....
Did not read the whole thread there, and the pot was in a can on the desk here.....
Put two and two together and got 3...hi...
They were talking about virtual basic er software on-screen operators....
gotta go now, I'll be back later.
still no luck with cvs...
looks like I did get your key in there right, I just checked
nope, my screwup, try again
yes! thanks, now the dev checkout works.
hmmm, how do I create a new directory with cvs ?
03awallin 07HEAD * 10emc2/src/hal/drivers/m5i20/hostmot5_4/hostmot5_4.bit: this is the bugfixed hostmot-4 with 33MHz pwm clock
03awallin 07HEAD * 10emc2/src/hal/drivers/m5i20/hostmot5_4/REGMAP4E: register map for hostmot-4
cradek: I need developer privs. What do I need to send you and where should I send it?
lerman: you need to create an ssh key pair
ssh-keygen -t rsa
then you email the .pub to chris
I saw the message telling you to do that. OK. I need an email for chris.
chris at timeguy dot com
03awallin 07HEAD * 10emc2/src/hal/drivers/m5i20/hostmot5_4eh/hostmot5_4eh.bit: this is the new hostmot-4 config with 100MHz pwm clock
OK. That's been sent off to cradek.
03awallin 07HEAD * 10emc2/src/hal/drivers/m5i20/hostmot5_8/hostmot5_8.bit: new 8-axis hostmot-8 with 33MHz pwm clock
awallin: I gather that you're in about the same position as I am. I checked out some stuff as anon, I've modified it, and now want to commit it back. Can I just do that from the directory I started with (the anon one)? Or am I going to have to create a new release, put in the same mods, and then commit that?
03awallin 07HEAD * 10emc2/src/hal/drivers/m5i20/hostmot5_8eh/hostmot5_8eh.bit: 8-axis hostmot-8 with 100 MHz pwm clock
03awallin 07HEAD * 10emc2/src/hal/drivers/m5i20/hostmot5_src/ (33 files): these are the vhdl source files for the hostmot-4 and hostmot-8 fpga configs
03awallin 07HEAD * 10emc2/src/hal/drivers/m5i20/README.txt: info about the different configs and how to use them
lerman: you can only commit from a devel checkout
so you need to do a new checkout with <username>@cvs.linuxcnc.....
and then commit from there
you can probably do a diff/patch between your anon and devel checkouts to save some manual typing
anyone know where a more detailed TODO list is kept ? or should I just add stuff to /src/TODO
But can I fool cvs by changing the CVS/Root file? Since all of my changes are in a single directory, I should be able to just copy the Root file from the new one to the anon one and then check in my files.
copy the modified files into the non-anonymous checkout dir
SWPadnos_ is now known as SWPadnos
Yes. But I have to do those one at a time.... Can you say lazy? And I have to know which were modified (yes cvs diff).
your mods should be only in the rs2764ngc dir, right? (or at least most of them)
awallin: whee, thanks for your work on this mesa stuff
copying the CVS/ dir is equivalent to copying the files, either is a little dangerous if you didn't start from a fresh checkout
Not sure what the correct etiquette is here. Can anyone jump in with their "opportunities", or is that impolite?
I've got a version with name parameters ready to commit.
it's not impolite
03awallin 07HEAD * 10emc2/src/hal/drivers/m5i20/README.txt: further info on m5i20 configs
cvs ci <list of filenames>
We all talk at once on many topics. It's not impolite, but you may get ignored by people who are thinking about something else.
Is it cvs ci or cvs commit?
when the editor pops up, enter a nice descriptive comment (use several lines if you need to), and save the file
ci is the same as commit, I think
checkin or commit
ci = checkin
jepler: thanks for noticing! I don't have much time for emc now but I try to do something...
I need to get back on the developers list.
ok - time to run again
Shades of RCS.
CVS = RCS, for very small values of CVS ;)
I am having "opportunities" with setting a thread in HAL. Sometimes it will accept the following command, and sometimes it will come back with a illegal character error msg.
gsj: sorry I can't help. I run emc1 on my real hardware and my only emc2 experience is with simulation.
"halcmd loadrt threads name1=base-thread fp1=0 period1=50000 \ name2=servo-thread period2=1000000"
Does it tell you what character it is? Or the position on the line? Or anything useful?
what's the backslash ('\') for ?
I assume to split the line. Are you using the '\' in the command you are giving? Is that the problem?
No, just that is has encountered an illegal character. I've tried the -v and -V options, but it didn't change anything. I don't know what the backslash is for. It is just like the one shown in the EMC2 user manual.
That is probably the problem. Type the whole thing on one line without the '\'. That was probably in there because the command wouldn't fit on one line in the manual.
OH, I thought it might be a delineater between the definitions of the 2 threads.
goslowjimbo: do it all on one line without the \
In Unix commandline shells, \ as the last character on a line means the command continues on the next line -- but it's often best just to take the simple approach and write it all on one line
also, "fpX=0" or "fpX=1" will not work if you are using emc 2.0.x, that was added later but somehow crept into the 2.0.x documentation
what does fp do ??
awallin: it specifies whether that thread can use floating-point calculations or not
you can specify 'fp1=0' to turn off floating point for that thread, and maybe make it take a microsecond less to execute
03awallin 07HEAD * 10emc2/TODO: added hal_m5i20 driver to-do list
OK, so then you can't hook up any floating point signals/operations or anything to that thread ?
how do you eliminate the floating point for emc 2.0.x?
goslowjimbo: leave fp1= and fp2= off the loadrt line and don't worry about it
awallin: each function must specify whether it uses FP or not. If it says it uses FP it can't be put in a no-fp thread
awallin: if the thread really does use FP, but says it doesn't, and you add it, you get an error in your system log and emc stops working until you restart it
SWPadnos_ is now known as SWPadnos
jepler: are you still there ?
I am wondering about the 3D view in AXIS
It's done in OpenGL ? but are you using some kind of toolkit on top of OpenGL ? or are you just using the standard OpenGL commands to draw everything ?
is it fast enough when implemented in Python, or is it written in some faster language ?
awallin: I use my own wrapper of opengl
awallin: drawing the preview plot is just one OpenGL call: glDrawList or something like that
some other parts of the drawing are done in C
it has evolved over time
I'd just like to have something which can plot stuff (lines in 3D mostly)
but there are a lot of advanced toolkits around, I'm wondering if they are worth it
or if I should just stick to plain openGL
one of the "must have"s for AXIS was working within a Tk application
Hi, I am having trouble configuring EMC to run python scripts
huggyb: the stuff described in "Program Filters" in the online manual (e.g., http://linuxcnc.org/docs/devel/html/gui/axis/)
wis a new feature that will be in emc2.1, but isn't in emc 2.0.x.
huggyb: so if that's what you're trying to do, that's probably why it's not working. but if you're using the development version, or that's not what you meant, I'll try to help you out.
in emc 2.0.5/axis 1.4a0, a much more limited "filter" was supported -- it wasn't very useful, really
That explains it. I have the ubuntu sruff installed
Maybe I should try the development build, thank you.
you can compile the development version yourself -- instructions are on our wiki
there isn't a prepackaged version yet
are you looking for developers?
yes -- It would be crazy to turn down a developer!
that's because you don't know me ;-)
do you have some ideas about what you'd like to change/improve in emc2?
Not really... I'm a simple C++/Java guy that got a sherline mill for xmas
I want to learn python doing wizards and other gcode generation stuff
are you familiar with stuff like CVS then?
read around on this page: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#Getting_the_source_with_CVS
actually you should start at the top of section 3
once you've compiled the CVS version (either v2_1_branch or HEAD) you'll be able to use [FILTER] stuff like the manual shows
I'll be in and out this afternoon, but if you get stuck just ask for help again
lerneaen_hydra is now known as Lerneaen_Hydra
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2006-12-30.txt
Lerneaen_Hydra is now known as lerneaen_hydra
off to Slab City for new year's.. see ya alls
making some brandy slush right now..
what a waste of brandy =)
skunkworks: 500cc of Brandy IV Push =)
Jymmmm is now known as Jymmm
If anyone has any ideas... My Sony TV no longer has color on the tuner or video inputs, but does have color on the OSD. I cracked the back cover on it, but I didn't see any loose connectors. Any Ideas?
no clue - sorry
subject of email - At warm in voltmeter
first line of emal - You got this message, so you lucky man. It is your chance!!
I'd try to reseat the modular boards, but not sure how to safely discharge the flyback
Anyone seen cradek recently? I need to get added to the cvs developers list. I did send him email.
I just wrote a test program for my changes. It 'mills' a flowsnake.
"Just say no to cradek!"
(And hope he doesn't say no to me.)
I'm on a roll. I'd like to get those changes in, so I can add some more.
[22:06:25] <lerman> http://www.meden.demon.co.uk/Fractals/flowsnake.html
It is the first gcode I've written that uses recursion. It's fun -- if not very useful.
Anyone know if there is a packgage to install Qt on ubuntu using apt-get? and what the name might be?
apt-cache search libqt
I just couldn't resist: http://lerneaenhydra.shacknet.nu/index.php?option=com_rsgallery2&Itemid=32&page=inline&id=16&catid=3&limitstart=0
cradek: did you get the email with my ssh public key? I'm hot to get my tuff out there.
lerneaen_hydra: I thought you already had commit access, but I see now you don't
I'll do it now, one sec
cradek: I take it you mean lerman ?
I used to. With a password. That was before the change to ssh.
oops, yes, lerman
hmm, link should be working now
l[tab] gives me one of you two at random :-)
lerman: give it a shot now
looks like it works
Yeah. It's doing something.
I am trying to get the torus png sample working on my machine but I get an "Exit code 1" when I load the file. Can anyone help? I just compiled the latest emc from cvs.
did you try it with the sim/axis configuration?
do you get the image-to-gcode window where you can configure the settings?
any errors on stdout/stderr?
I can get holecircle.py to work
I get "No module named numarray"
Is HEAD the default when I co? If not, what do I do to get HEAD?
you should install numarray
huggyb: sudo apt-get install python2.4-numarray
lerman: funny photo!
I also need ImageTk... do you know the full package name?
lerneaen_hydra: funny photo!
cradek: the wonderous days of highschool ;)
Its working now, thanks :-)
huggyb: I still haven't used that to mill anything - let us know how it works if you do
When I use any of the python scripts, how can I save the gcode? cut & paste don't seem to work from the console
you can run the python program at the shell and redirect its output to a file
great. This is cool stuff, by the way
glad to hear that
guys, i have a question...
about this powersupply design, http://www.kg-uitdaging.nl/COlabvoeding/CO%20labvoeding%20V2%20schema.pdf
Bo^Dick: might be better to ask on #electronics
Does anyone know what the ground pins are on a parrallel port? The halscope says I'm not getting into the computer, but I'm driving it with a encoder. The ground pin is the only other thing I can think of.
within minutes of me uploading images to my site there are hits on the images O.o
lerneaen_hydra: I'm reading your webpage - I didn't know you got two of those lathes, and for free, that's very cool
cradek: oh, that explains a bit
they're pretty good, aren't they? ballscrews?
cradek: yes, I really lucked out
cradek: they are ballscrewed, yes
very nice find
how's the latency of the site?
it seems fine
(it's just hosted on a computer at home)
The requested URL /images/files_projects/compact5/img_5238.jpg was not found on this server.
I think all the emco images are broken
I just moved them over to the gallery
haven't fixed the article yet
did you ever get threading on that lathe?
I remember it had the sensor already
the article should be fixed now
yeah, threading works now
quite well actually :)
using the counter module?
you should make a new video :-)
ooh, that would be nice
I'll have to do that soon
then I can add a video to the gallery too :)
hmm your AXIS is not in lathe mode
or maybe the photo is old (before there was a lathe mode)
forget it - the next photo is
yeah, that's an old image
03lerman 07HEAD * 10emc2/nc_files/flowsnake.ngc: Add a sample program using named parameters.
[23:11:14] <cradek> http://lerneaenhydra.shacknet.nu/images/rsgallery/original/img_5398.jpg
I've even updated the images now ;)
03lerman 07HEAD * 10emc2/src/emc/rs274ngc/ (8 files):
Add support for named parameters. Should not break any existing
Try it, you'll like it. The flowsnake is pretty cool -- although I wouldn't want to try machining it. Engraving it would work, though.
joomla is really nice to work with after you've adjusted to it
looks nice and doesn't take too much work
flowsnake works here
That was quick. Of course that one works. If you have anything you can use for regression testing, it would be appreciated.
heh, one of those fractal triangles would be cool
useful-subroutines.ngc works too - I think that's the most new interp features I've ever used
lerman: it's strange (but correct) to see it do the whole shape with just line 13 highlighted
Why is that correct?
because line 13 is the G1 that causes the motion
works here also :)
I'd rather see it step through the calls and everything whether it does motion or not.
lerman: the interpreting and the motion happen at very different times. I think having it highlight as it reads ahead would not be very useful
Useful, I think, but differently useful. It would be nice to be able to single step through the code to debug it.
From the user (machine operator) standpoint I think it does what it should.
I'm going to take a look at debug comments. (debug, a param #1 another param #$foo$)
lerneaen_hydra: do you know martin persson?
cradek: heh, the distribution of hits on my articles is interesting
One issue I has is where do we output this? If we do it like msg, it requires an ack for each output. That would be a pain.
using single step on flowsnake has a surprising result
fenn: I recognise the name
he's a staff member at chalmers i think
anyway that's neat
(well, not surprising to me - I knew what it would do, which is why I tried it)
he is xarragon on freenode
fenn: I don't think I've met him
ooh, haha, cool
How do you get it to single step? Once you hit run, it starts. Do you have to stop it and then single step? Then you couldn't step from the beginning.
you have to run, then pause, then step
fenn: oh, now that I look at what he's done I think he may have held a presentation/lecture when we started about global warming
right.. he says he's working on doing dsp algorithms in an fpga right now
spiro program works also
Watch that first step; it's a lulu. :-)
fenn: oh, cool
skunkworks: that's good to know, it was pretty complex too
has etla changed his name?
hey, fenn wait a sec, how do you know martin?
we were making fun of bo^dick in ##microcontrollers last night
is he often there?
hm.. not sure, but i see him in ##electronics from time to time
which channel do you guys use to make fun of me?
I use #jepler-has-a-funny-hat
-!- Irssi: #jepler-has-a-funny-hat: Total of 1 nicks [1 ops, 0 halfops, 0 voices, 0 normal]
you guys are so mean :)
I like it
sorry jepler - I know you like your funny hat :)
what is this hat you're talking about?
it's in my mind, I think
Again-- looking at: 'm going to take a look at debug comments. (debug, a param #1 another param #$foo$)
One issue I has is where do we output this? If we do it like msg, it requires an ack for each output. That would be a pain.
cradek, jepler, SWPadnos, any thoughts on this?
lerman: I think stderr, at least for now
I could just output it to a log.
I think it should be identical to the current msg facility
Where does stderr go (these days) -- in axis?
stderr is stderr, it goes to the controlling terminal unless it's redirected somewhere else
and there isn't one if you run from an icon (is there?)
I guess it would go to the X console
Well, the neatest thing would be if axis added a third page (manual control, mdi, LOG).
there's already a way of handling messaged - I'd just improve that by adding formatted parameters to it - don't change how it works though
stderr get us nothing. I'd rather create a log file and output to that. At least you could then tail -f it to see the results.
SWPadnos: currently you have to click OK after each one
The problem I see is that each msg requires an OK (cradek is faster).
if that's what a message is supposed to do, then ones with parameters in them shouldn't be any different ...
that's an interesting point.
you are, after all, messaging the operator
ok I agree with SWPadnos
maybe there should be another command added: (log, blah blah)
Well, for now I'll write the code like that. I notice that the MESSAGE function calls some python stuff.
and "print" - to go to stdout/stderr
in the interp?
It really isn't for the operator (although it could be used for that). Once the code is there to insert params, we could have several versions.
lerman: you're just going to call MESSAGE() right?
For now; yes.
lerman: ok, you could make another canon call later.
I think if you want to add some logging or error capcability, that shouldn't be a nsg - it should be called something else
msg, that is
nsg would work, but nobody would understand it :)
I think print or printf - and it goes to stderr.
I wasn't calling it msg -- msgf, or debug. Calling it msg would break any existing messages that had the character '#' in it.
I didn't think of that
ah - ok. debug to stderr and print/printf to console
To console == msg with formatting?
unless the code looks at the #xxx token, and outputs it unchanged if there's no xxx defined ...
sure, anything you're adding can have formatting built in from the start
But there might be one defined?
but you have a good point about breakage with existing msgs
^H over the ?
sure - in that case you'd get something funny looking
something like "change to tool #47" could be an issue :)
Of course, if someone has a comment that looks like (debug, whatever) I've already broken it. But my view is that it's pretty unlikely. Rumley had added (in emc1) the ability to have simple .ini file stuff. I was thinking that an ini file entry saying "USE_EXPANDED_COMMENTS" would at least allow the inegrator to guarantee compatibility here.
Yes SWPadnos -- that was exactly the problem I had in mind.
bb tomorrow or so