how strange. I just watched a couple of interesting political debates
I never knew there was an "Impeach Bush" party :)
what does it take to update to 2.0.4
just a run of the update manager
like updating any other package
it looks like the update affects steppers only
chris what is the wrench looking tool next to your name on the developers list/
you can click those to find out what they mean - it means I am an administrator of the project
is that why it's followed with a dollar sign cause you make the big bucks.
I don't see why you think the update only affects steppers. the changelog only mentions steppers in one item
I've never had anyone donate money to me... I don't think people really do that.
maybe I should turn it off.
i wouldn't do that, it makes you standout in the crowd.
hmm, I can't figure out how to turn it off anyway
what does the developer entail.
whazt do you need to do as a developer.
participate in improving the software
what parts of improving the software are needed. is this piece meal or is there some dorection.
the board of directors sets the general direction of the project, but developers are fairly free to do what they're interested in, as long as they work together with the whole group
often several of us work together on a complex change
my interest lays in cl....i like to find those of same.
I have to run, bye
Anyone heard of a project that runs a full linux install on a USB thingy, that when plugged into a M$ box it opens an x session terminal to the linux install on the USB thingy?
steve_stallings is now known as steves_logging
theres video too
under 'view demo'
I've started work on my Gcode Wizard. You can see my first screen shot at:
[01:53:45] <lerman> http://www.se-ltd.com/~lerman/gallery/CNC_conversion/wizarddemo?full=1
Comments, please. Ken
oooohhh - Elrod stuff
I wish I could afford their motor mounts
Yup. Not cheap, but very nice.
Elrod machine - they make very nice CNC cpmponents and kits
I think I asked Dwayne about "scratch-n-dent" goods, but he still didn't have anything in my price range :)
lerman, that's a good concept. I'm not sure how the screens should be defined though
I was anxious to get my machine running, so I just bought the stuff. I'm using Jon Elson's boards.
it seems that bitmaps with edit boxes in adjustable locations is good, but I'm not sure it's generic enough
yeah - I have a USC as well
That screen has been defined as a background file plus a configuration file.
I snagged an Anilam quill drive off eBay for $300 or so, but the X/Y mounts are still $1900+
I do have a combo box. And I plan to add radio buttons.
I waqs thinking of the graphics though - in the example, the spiral direction could changes depending on the climb/conventional setting
I more or less copied the content of the screen from the newfangled addons for mach3.
SWPadnos -- are you really there or am I talking to a computer: "right, yep, sure". I could write a program that did that.
you must have missed the comment about spiral direction ;)
I can see how it would get loost after a "yep" though ;)
damn - I think all this painting triggered a nosebleed - brb
Yup. The climb, versus normal is supposed to take care of that. The mach3 stuff also has a cutter direction CW or CCW. I don't think it's needed. Also, I'm assuming that speeds and feeds are set for the job in another screen.
And I'm not sure we really want ramp angle. I think we really want to specify the tangent of the ramp angle. -- that would be specifying the slope.
Also, the mach3 stuff doesn't specify finish cuts.
actually, CW/CCW is needed - there are left-hand spiral cutters
I've contemplated having the ability to specify a validate program for each. It could be a program or script that processed the args and validate things. It would see if things made sense. A finish cut that is larger than the step depth,wouldn't.
I'd do that in the wizard if possible - limit the inputs to values that are consistent
Well, it is easy enough to add that to the screen. It just adds another level of complication to the gcode subroutine that would run.
that gets difficult when things are interdependent though
how are you looking at these on the back end?
Well, the wizard itself doesn't know about anything. It presents a screen and generates a call to a gcode subroutine.
are you using templates that get filled in with mathematical formula results?
There is no math. Or rather, the math is done in the gcode subroutines.
ah, so the subroutine has all the smarts, the G-code produced by the wizard is a single line with a subroutine call, and possibly several lines to set up parameters
That was part of my motivation for building the subroutine stuff.
can we include G-code files yet?
The Onnnn call is a single line that has all of the parameters.
(ie, is there an include directive in the interpreter)
No, not yet. For now, this thing would insert the routines in the beginning of the file.
I assume the interpreter is single-pass, so subroutines need to be defined before they're called?
The middle horizontal pane is intended to display the entire gcode file (scrolling). If a line is selected, it is displayed either as a screen (if it is one of our calls), or in the bottom line editor where it can be changed.
Yes. It is single pass (it is a real interpreter).
This thing will let you edit a gcode file that is a combination of wizard magic and hand coded lines.
sure - that's cool
does it handle subroutines that weren't generated by wizard screens?
Its written in c++ using Qt.
(ie, show them as editable lines)
Sure. Just include them and they are editable. By include, I mean insert them manually.
ok - I was wondering (as a programmer) how it can tell the difference between Onnn calls that are "auto" and ones that were manually written
But adding a wizard screen is very easy. The hardest part is the graphics and positioning the stuff. I'll probably add a tool to do the positioning.
I was thinking of a language helper like comp
The auto ones are ones that are calls to subroutines that are in the library. I plan to add a configuration thingy to let you set multiple library trees.
What is comp?
it's a tool that eliminates lots of the boilerplate from writing HAL components
look in emc2/src/hal/components, I think
Ah. Yes. I remember seeing it.
This would be another gui thing that let you put up a screen and then add the widgets corresponding to each argument to the gcode routine. It would automatically generate the config file.
I'm thinking of something that you can stuff G-code snippets into as well, with some way of defining how the "screen variables" affect the subroutine (if they do_)
It would sure be nice to have a Qt guru to collaborate with on this thing. I've had a hell of a job trying to get the panes sized properly.
heh - that should be the easiest part ;)
dealing with anchors and autosizing is a pain though (in the limited experience I have with Qt)
Well, the screen variables are arguments to the subroutine invocation. The config file has them in order.
you get magically increasing text when you don't want it
you're limiting it to something that can only do a subroutine call
I'd allow for expanding things into G-code macros as well, for things like a "job setup wizard"
Correct. That's because everything can be done in a call.
that could have inch/mm, for example, and tool selection ...
I'm not sure that we have macros.
Just write a subroutine that does the tool selection.
sure, but having a subroutine that gets called as O3000  that then does G#1 (G21 in this case) is ugly as hell
inch/mm is a bit of a problem in the sense that I might want to vary the resolution in the gui.
the macro idea is that the g-code generator should be able to have screens that can generate actiual G-code, not limit it to just subroutine calls
Ugly, but the user really doesn't see it. He just sees a tool selection screen.
That would require some smarts in the wizard. -- or perhaps an additional or different abstraction.
that's why I'm mentioning it now ;)
actually, most things could be done with "alternates" - ie, the wizard writer provides a different piece of code for each item in a drop-down, for example
for other things, some math machinations would provide answers based on other parameters
other things would be essentially variable substitutions
One big issue is cutter compensation. Clearly we could write circle routines that do that, but things like the stepover are dependent on the tool diameter. Is there a direct way for code to read the diameter of the current tool.
I'm not sure there's much else
I don't think so, but I'm not sure
I think you need to write code that assumes a particular tool - like the one you provided that asks the user
you can ask the cutter size/shape, and then calculate astepover from that (by having a "scallop depth" edit box, for example)
The wizard writer could provide separate code, but my view is that the gcode would have an IF at the beginning to select the right part.
well, that works
One problem that we might have is name space issues. We can support multiple circular pocket routines, but every subroutine must have a unique Onnnn. Someone has to assign those.
yeah - that's where some flexibility in writing the g-code would come in handy
having a large circular pocket subroutine that can do every type of curcular pocket imaginable is great, but f it makes it unreadable or unsuable without the wizard, I'm not sure it's the best plan
I did contemplate adding something to the gcode definition that would let one specify a starting Onnnn at the beginning of the subroutine and everyting after that would be offset by that value (on the fly). We could do that as part of the include facitility. Include filename Ostart. Then all of the code would be written with a zero base that would alway be added.
sure, but the other code needs to know the new Onnn "name"
this is where named subroutines would come in handy
Well, right now there are 13 args. As long as the user has a list of them he can use it. I've sometimes written routines like that and then nested them inside one with a simpler calling sequence.
like a C++ family of constructors approach
I did consider adding that to the language.
yep - I remember a short discussion recently
Named subroutines doesn't solve the problem. You and I might each have our own CircularPocket routine. In some way, it make it worse because clashes are more likely. It makes it easier because at least the clashes are visible and obvious.
and there's something relatively unique to search for
But please don't use a label called LOOP.
about tool diameter - would it make sense for the user to specify the tool in a tool table? might be good for people with tool changers
By convention, we would probably want everything in a package to start with the same string.
Can we read the current tool? And get the values associated with it (from within gcode)?
jtr, that makes sense for a machine, but I think it would be great if the wizard could be run elsewhere as well
I don't think you can get the diameter with G-code
it could probably be made available as a parameter though
(it may already be - I doin't know)
That's just plain dumb. But we could easily do ... what you just said...
Right now, the wizard can display and let the user modify things. I think my next step is to read out the values and generate the single line of gcode. Then its on to displaying multiple lines and editing them.
I think the code substitution idea I have could be added later
so getting a good UI framework is still the first order of business
argh. I've got to go. the paint fumes are about to kill me
Did you realize that you screwed up when you mentioned your Qt experience? I think you just acquired the job of reviewing my code. My stuff is going to look like a pretty raw mix of C and C++.
Yes. UI is first.
no problem. I'll look, but I can't guarantee that I'll spot any errors ;)
Thanks for the feedback. Well talk again later, I'm sure. I'm off.
[02:59:32] <skunkworks> http://www.electronicsam.com/images/KandT/servostart/permotor.JPG
wow that's what I call full specs on a motor
with units even
50 megaseconds at 150A - not bad ;)
if it's american, it might be 50,000 seconds because M is the roman numeral for 1,000
yep - Mille seconds
then again, MFD is usually microfarads
at least we don't use that confusing metric system
but 50Mohm on an old schematic is 50,000
right - I've also seen MM used for million, usually in financial reports (sales of 2.7MM / year)
ah I've also seen MMFD for e-12
never seen that one
really, it's common
hope I never do
probably for old radios??
[03:46:16] <cradek> http://www.justradios.com/MFMMFD.html
it's a good thing they didn't have 10 GHz radios back then
you'd have inverse roman numberals 12 characters long
[03:48:59] <cradek> http://www.philcoradio.com/tech/resistor.htm
here's a new one for me, the body-tip-dot scheme on striped resistors
roman numeral radios?
yeah - I've never seen it before
"so tune to station XCVIII point VII for all the latest hits"
I know I've seen body-tip-dot colored resistors but I had no idea what the code was
yeah, the dot, which is in the center of the wirewound resistor that always did run a little hot, is always a "mud" color
maybe using colored stripes instead of printing numbers is a bad idea.
yeah - think of the color-blind
lots of males are red/green color blind
I know someone who used to work with a color-blind electronics tech (male), it was a constant problem
I guess he had one colored contact lens that was supposed to help, but didn't
hmmm - maybe it would turn red or green into black or something
yeah I bet it was something like that
a colored contact lens? Interesting -- if you could learn to see that way, I guess it might overcome colorblindness. I hear the human visual system is pretty adaptible.
yep - partially inverted fields righting themselves and such
yeah it seems possible, not sure if that was his own cleverness or if it was/is a standard treatment
Julia Sweeney has a funny piece about the eye on her "Letting Go Of God" CD
(it's on her website - I don't have the CD)
was she popular before she ditched her religion? I don't quite get why she's so well known for doing it
[03:57:07] <jepler> http://www.martymodell.com/colorblindness/fdatalkpaper.html
she used to write for Saturday Night Live, among other things
she created "It's Pat", I think
she may have acted on the show as well
[13:46:54] <skunkworks> http://www.electronicsam.com/images/KandT/servostart/permotor.JPG
why do you keep saying that?
sorry - crashed last night. I read yours and cradeks comments on it ;)
skunkworks: cute little motor
skunkworks: bet you won't impress JMK with that one?
I have no clue where it came from - it has very little inertia compared to the other ones.
alex_joni: I doubt it - but it will be a nice size to play with.... work my way up to the moster ones.
Father just pulled it off a shelf - 'What about this one" he says ;)
some times its nice to have a horder in the family
SWPadnos: the encoder wheel and sensor looks nice.
that's quite a cute little motor actually
concidering it can only do 18 in-lbs continuous
argh, use real units >.<
and 300 in-lbs max (for 50ms) ;)
Lerneaen_Hydra: that's 288 oz-in
2.5 pounds in a kilogram?
$ units "18 lb-in force" "newton meter"
2.03372692 N-m, according to google
oh, I guess my estimate of 2.5 wasn't too far off then ;)
1.8 Nm isn't a lot ;)
but good for experimenting :)
but the angular speed was quite good (366 rad/s)
~270 oz inch..
about 2/3 the size of a servo for a Bridgeport-sized machine
it is physically pretty big - about a foot long and maybe 5 inches in diameter.
what is it, an asynchronous motor?
they must have sucked at designing motors back then
(judging from the fact that my motors are 27 in-lb, with peaks at ~110 in-lb)
my small servos has 300oz-inch of peak..
permament magnets it says
so servo motor?
and they are small
anonimasu: what CAD do you use at work?
03jepler 07HEAD * 10emc2/src/Makefile: copy bitops header file into place
somebody wants to explain me about gantry ?
Be more specific. what do you want to know?
i have 2 motors for x
what to do in hal ?
Ah - I know they have been working on a solution to ganged motion.. (I am not going to be much help)
But someone here should be able to answer your question
* skunkworks pokes alex
* alex_joni pokes back
did you read the recent discussion in the mailing list archives?
well guest764 just left..
is it "linksp Xpos-cmd => pid.2.command" ?
yes i have read it
do you know how you are going to home?
you cannot just hook this up without understanding it
yes there is 2 sensors
in first time i want just move 2 axis themselves
initialization is second time
something wrong with the question ?
if you want to move them as 2 axes, just define them as 2 axes
I'm not sure if it will work with g-code .. but you should be able to jog them
but jogging them individually can literally break the machine
it's a beginning
you may need to order them as XYZX though
I don't understand this answer
both of yours
XYZX would map to j0..j3
cradek: basicly I was saying to set it up as 2 joints
right - in other words, the first 3 axes need to be XYZ, I think
DANGEROUS but ok for testing
how can you home?
19:23 < Didier> in first time i want just move 2 axis themselves
19:23 < Didier> initialization is second time
oh I didn't understand that I guess
USE_HOME_SWITCH = NO
I think you want to use something like jepler's parallelkins, or set up something in HAL for individual control
hit home on both
it's too fast for me
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2006-11-06.txt
Didier: you can read it with no hurry there .. :)
i will try
you could use parallelkins with manual homing, but if you want real homing I think you will want to have a complex hal setup and trivkins
and maybe some external hardware...?
jon elson explained how the homing process has to work
seems to me that setting up the two Xs as separate joints is dangerous
move both until one hits its limit switch, stop that one, and keep moving the other until it hits home. save the difference, and use HAL to apply the difference to all position commands from then on
or something like that
but it requires a bit of rewrite of the software
i can change the software
or you could do it in hardware: each limit switch stops the opposite joint, and when you trigger them both you tell emc home has been hit
I think jon E said something like that
err - you can never trigger both if one limit switch stops the other joint
since hte other now can't move onto its limit
err you know what I mean :-)
each switch stops its own joint
ok - that makes sense :)
Didier: what I'm saying is you can't just ask us what hal command to type. You have to understand what EMC does, then design your scheme, then implement it in hardware or HAL
_and_ adapt a few bits of the software (e.g. EMC)
maybe, or maybe you won't need to
I somehow have my doubts you can do this only in HAL or HW
but I've been proven wrong before :)
I think it's possible, but it would be ... intricate
do you think : It's possible ? If yes, it's easy to do ?
you could align the gantry by hand, or home on one side and measure the offset and apply it with halvcp
Didier: yes, no
thank you cradek
Didier: you can do it easy, but then it's not 100% ok
Didier: or you can do it ok, but then it's not easy
and if you do it without full understanding, you can break your machine
and emc ;)
it's easy, if i use only 1 home switch ?
Didier: it's easy.. but then you won't know if one axis is out of sync
yes but you have no provision for alignment then
i understand well, in that case there is no alignment
in that case, only XXYZ
Didier: you need to align by hand and hope it will not lose alignment
no, in that case you only have XYZ
and the X->X'+X" in HAL or hardware
somebody have done yet ?
at least one
Till Franitza responded on the list
but again.. if you have problems with the alignment.. it's bad
i know, the quality of cutting...
no, the machine destroying
if one side stops and the other one moves.. then you can damage it really bad
no following possible to stop the machine ?
i just want a master/slave
ther will be a following error which you can use
but once it errored .. I'm not sure how you can make it straight again
i see the problem
admital is always the first step
i have to change the hal
there is no example ?
Didier: none that I know of
but if you read tha HAL Documentation, and get used to it you can probably do it
well, i try
thank you :)
i'm sure, i will come back
talk you later, bye
thank you everybody
cvs compiled version vs dev downloaded source code? i have on this machine a cvs version set up for stg.i also have a dev.pkg . any precautions to look at before making another emc2.
mtedad: not sure I understand the question
emc2 is compiled and runable, cvs version. i want to do dev work with downloaded pkg with all the sharing of files in linux how do you differentiate between 2 version
you put them in 2 different folders
alex while your there the file emc1.in emc2-2.0.3/docs/man/man1 looks to me as parameters used with "emc" in terminal to start emc.
well, just like everything else except emc i can't get them to work.
emc1.in is not parsed
after you run configure you will see a file called emc1
looking for theline to follow that will lead me to the culpritthat's keeping me from running all the little things in the manual.
have not configured the second emc. found this reading thru the download files. tried the commands on tyhe running version.
you need to issue commands like bin/halcmd and scripts/emc
maybe that's why it's not working for you
bin/halcmd or scripts/emc come back no such file or directory
scripts/realtime start don't work either.
ls /bin works
not /bin bin/
don't understand that
you said ls /bin
you need ls bin/
ls bin/ returns no such file or directory.
don't understand that
try that command and see what it says
took me home/mtedad
where did you put the emc2 code?
mtedad: I suggest looking around for a basic linux tutorial
i did nit put the emc2 code anywhere .cvs put the running code someplace. the downloaded files in sub dir off home.
yeah, and you need to be in that dir
the running dir?
give me the name of a file that would only be found in the running dir.
VERSION only found in the dir i was in.
mtedad: sorry.. I need to run for a while
skunkworks: maybe we should spam #emc not -devel :D
spam spam spam spam spam!!!
yeah.. gum wrappers
[19:45:42] <skunkworks> http://www.youtube.com/watch?v=MHHyggzoNDk
'see colon slash windows slash greater-than.'
aww - you can't see it?
come do think of it, are induction brakes ever used? (strong magnet + aluminum plate)
s/magnet/magnets or electromagnet
Did any of you see the discussion I had with SWPadnos, yesterday? (aside from SWP)
about the conversional programming?
* alex_joni didn't
lerman: kidding :D
I read it hastly this morning
I mean suggestions, comments, etc.
lerman: I'm not sure
I mean I never used conversational programming..
One thing I worry about when I build a tool is whether I'm building something that isn't usable, or is of interest only to me.
but I've seen a few commercial systems
and they had something like a different levels of menus
with graphical displays of what screens you would get
am I making sense?
Is it sufficient to have a tree that says milling operations, pocketing, etc?
* alex_joni is thinking touch screen interface
The mach3 newfangled interface has an overview with a button for each.
How would you enter numbers with a touch sreen?
* alex_joni doesn't give a s... about mach3 :)
But mach3 has a large user base -- because it is apparently usable.
I think mainly because it's commercial
this thing costs money.. must be better than the free one
But not a lot of money.
Touch screens for a shop environment are pretty expensive.
My style is to sit in the office, do my design, and then go out to the shop to run it.
lerman: that works too
I have one of those rubber keyboards that is shop proof. My display is a Xerox lcd display with a heavy GLASS front. I'd hate to think what a shop would do with one of the cheap displays.
there are touch screens which you can fit behind lexane :D
At any rate, a major goal of my 'wizard' is that any old user can add new screens and operations to it -- without having to be a computer programmer (but must be a gcode programmer).
I can see that
I'm sure I could build a touchscreen interface to the software. But, since every device is a little different, we get back to the system integration problem(s).
what do you mean?
A system integrator would have to make it work with a specific type of touch screen.
I would think the touch screen would just be a virtual mouse.
But you still have to be able to enter numbers.
lerman: I use a teach pendant on my robots
they have a 8" touch screen
and for number input they popup a virtual keypad
I'm sure virtual keyboards/pads exist under linux
Once you select one of the fields, how would you enter the center coordinate of a circle.
that is how our sankio robot is also
lerman: select the field, pop virt keypad, enter number hit ok
Yes, I guess you could pop up a number pad next to the field.
select next field.. etc
lerman: you shouldn't worry about that now though
that's easily added afterwards too
OK. I guess that could be a later addition. Should be reasonably easy to do.
but something like the main screen from where to select screens is not as simple
What should I worry about new (besides the election ;_)
as an alternative to your treelike navigation on the left
Why is that?
maybe because it would involve an icon for the screens
and a way for automatic sorting? (guess the tree needs that too)
So, I would just add another file in each directory that contains the icon.
so each screen is a dir?
Did you grok that each 'screen' corresponds to a separate directory containing a screen, a config, etc.
shit, you type faster than I do.
* alex_joni tries to :D
And I've probably been doing it longer (typing I mean)
bet you have :D
* alex_joni only started 20 years ago :D
Of course, I'm sitting on the couch with a mini notebook in my lap.
I started about 55 yrs ago.
Nope 45 yrs.
I'm actually cursing at my keyboard now.. used to the laptop keyboard
this ones too big for my needs :D
I've got this sony vaio with undersized keys I'm using. Takes some getting used to.
still better than typing on a cell phone :D
* alex_joni did a lot of instant messaging through GPRS
a few years ago
Do you really think that icons onbuttons will be easier to use than a tree with text?
I know it has more pizzazz
"ooohh.. shiny" effect :)
hmm.. how about you make the dirs a tree structure?
My plan is that you will be able to setup a list of libraries to pick.
a dir can contain more dirs and a picture
I thought I did.
oh, so you don't have a folder called screens/ and lots of dirs inside?
I suppose I could include an icon of a mill on the line that says milling operations and one of a lathe on the line that says lathe operations. Then I could have an icon for pocketing .
yeah.. that's my thinking
Actually the folder name is not used at all. Inside the folder is a file called 'desc' that contains the one word and the summary and line descriptions. Thenthere are sub directory to make the tree.
The icons would have to be pretty small to fit in the structure I have. Or there could be multiple screen to display and traverse.
But that would be slower than being able to see a large section of tree at once (like you can, now).
yeah, for normal uses I guess tree is best
this somehow sounds like it could use a "better" config method than dirs and files
Will it be a problem maintaining the source with a whole bunch of files named screen? and a bunch name config and a bunch name desc.
not sure.. but it doesn't feel very modern/advanced :D
not that it's a bad thing
Yes. If I new xml, I might take a try at it. But you really need a separate file or directory for each module so that users can add their own.
Also, I'm hoping that others will 'publish' libraries.
you said you planned to add a wizard to add screens
I mean help develop screens
I'm sure Jon Elson has some stuff he can contribute.
Well that's down the road. All it would do is let you move widgets around a background screen to get them placed right.
and load a background image
It was a pita moving them around by changing the config file and seeing how it looked; then doing it again.
* alex_joni knows
been there done that
I used glade
and I cursed every moment afterwards I decided to do that
The position is given in fraction of the screen ht and width so that if the image is scaled, it has a chance of working.
I drew the screen (the ONE existing one) using Visio.
"it has a chance of working" sounds promising :D
It doesn't have a lot of pizzazz, but seems highly visible.
And I think it shows the parameters in a way that can be understood.
pizzazz == flash
a button-based tree thingie could work - the only difference is that you have to go up/down the tree to see all the options
piz·zazz or pi·zazz or piz·zaz Pronunciation (p-zz)
1. Dazzling style; flamboyance; flair.
2. Vigorous spirit; energy or excitement.
whereas a tree usually allows expansion of multiple sub-branches simultaneously
oh.. flamboyance.. why didn't you say that :D
extra pizzaz adds to the "ooooh ... shiny" effect ;)
Yes. But that would be slower. (traversing by buttons.)
yes, but fully compatible with (a) uncoordinated mouse users and (b) touchscreens
Speaking of something I absolutely hate -- if we want pizzazz, we could add.... skins.
fsck. breaking a nail hurts when it goes down several mm into living flesh...
actually, using gtk helps in that regard, since those apps should take on the current OS theme by default
if a suitably recent version of GTK is used
Of course, since this is user configurable; a user could change the screens to whatver he likes.
tip: don't get too hasty when filing ;)
Lerneaen_Hydra: pliers and rip it out
it'll grow back eventually
Not using gtk. I'm using Qt. (and in fact, I'm buiding under windoze for now). My linux systems are in my office an in my shop. I'm in the family room.
heh .. linux isn't family'ish enough :D
there's a Windows version of gtk as well
* alex_joni is kidding
yeah .. gimp uses it
and Gimpshop rulez :D
the gimp developers developed it ;)
Well, for day to day ops, my notebook is running windoze. I have two fixed machines one for me and one for my wife in the family room, too. Also my Asterisk telephone system. (that runs linux).
if ever you feel that you're having a stupid issue with linux... do to the #ubuntu channel, and you'll feel so much more proficient ;)
I probably qualify (or did once upon a time) as a unix/linux guru.
But there is so much new stuff that it is hard to keep up.
* Lerneaen_Hydra is not by any stretch of the imagination, but can at least do simple end-user stuff
that's true for most things :)
EMC would be a lot easier on Solaris than on linux. Solaris has real-time user space code.
but solaris hardware isn't as easy to aquire for cheap :p
not to mention running other apps ;)
do they still build hardware?
But it's interesting when you write a loop that is real time and hangs. -- lockiing out everything else.
I think they do
oh, apparently there is x86 support to
Solaris IS available on x86 hardware. Sun != Solaris.
here I was thinking it was only for sparc
lerman: right.. since 9 something iirc
java and RT doesn't seem to mix though
lerman: emc used to run on solaris
solaris realtime is unlikely to be good enough for a machine controller
Lerneaen_Hydra: there is RTJava
aren't those antonymns?
with a RT Garbage collector
(java and rt)
One trick I learned is that you write a RT app that is at the highest priority and can be used to kill the other RT apps. That way when your RT app hangs in a loop, you have a way to kill it without rebooting.
what is RT? what polling speed can you get?
I don't recall, anymore. But damn fast is what I remember. And it was HARD RT. It had a well defined maximum latency of a small number of microseconds. I'm thinking 10 to 40. But that was a loooong time ago, so I'm sure it is much faster, now.
oh, that's quite good actually
what was it used for? controlling mechanics?
ie very powerful PLC replacement?
possibly for communications streaming / mux and demux
I don't think they're ever had peripherals like motors, but high speed comms and audio/video they do have
We used it for communication. Not for a hard RT app.
I just tried googling to find solaris latency and didn't come up with an answer.
woo, anyone remember that electric bike I was ranting^W talking about before? I've done a bit now (motor and middle gear bit). ftp://basic:email@example.com/temp20061106/motor2.avi
==> PORT ... done. ==> RETR motor2.avi ...
Error in server response, closing control connection.
hmm, it's been flaky for a while now, just a sec
there, should be working
and apparently it is
is it working for anybody?
lerman: I have been playing with your oword stuff... great addition
according to my logs someone got it, with the same connection parameters as someone else who didn't manage to get it
* Lerneaen_Hydra does not like being stuck on a windows box
Thank you LawrenceG.
lerman: if you need a routine for your conversational programming project, I just did a boring subroutine
should get it
cradek: could you test with PORT instead of PASV?
it should be working though, as I've set up NAT correctly...
==> PORT ... done. ==> RETR motor2.avi ... done.
I think it is using PORT
and it dies?
it never seems to transfer anything
this does seem to work: wget --passive-ftp
==> PASV ... done. ==> RETR motor2.avi ... done.
cradek: are you 198.183.6.xxx?
oh, that was jepler
ETA35% [============> ] 703,084 22.60K/s ETA 00:54
yes that's me
we're both on the same network
both on that subnet seem to be working
* Lerneaen_Hydra shrugs
it does with with PASV
I'm going to wipe this installation soon anyway, so I can't be bothered to fix it
that looks like a pretty big motor to hang on a bicycle
it will stick out 2 or 3 times as far as the pedals, it looks like
it will be placed behind the saddle, over the rear wheel
warm your nuts
argh - looks like it's time to reboot his machine
LG: It's a bit premature, but thanks. That's one routine I have. One problem we will see is that it is not self evident what the appropriate functionality is. To ramp, or not to ramp. Allow CW or CCW cutters. Climb or normal milling. Where is the feed (and speed) specified. (Is it preset?, or set in the subroutine?) Does it have cutter diameter compensation, or not.
where the smaller unused gear goes to a 48 toothed gear on the starboard side of the rear tire
A nice thing is that anyone can write one of these AND add it to a library.
(so that long brass axle will be cut off)
lerman: yup.... I did some of the MOPs in the CP1 project... it was a real pain to code in tcl... I look forward to seeing your progress on a replacement
I suspect that we will have the typical Linux convention that there is more "stuff' than we can handle. But we will be able to add the libraries to a web page with descriptions, algorithms, etc and the user can just pick the one he wants.
lerman: the grapical display of what the params do will make it much easier to use
I guess one file that I will have to add to each wizard directory is a help file. That can include things like the author, algorithm, etc.
Yes. That's the idea.
I have some of my own libraries. Each time I use them, I have to look in the code to see what the args are.
This will solve that problem.
lerman: the oword subroutines have been very useful, but keeping track of them has been harder... debugging them is a bit of a bear without being able to display variables
I'd also like to add a facility to create new canned cycles. It just doesn't make sense to me that canned cycles are predefined and there is no way to add additional ones. I've written my own peck drilling code with a variety of flavors. I would add a configuration thing that says to map a Gword (say G83) to a particular Oword.
That's another thing on my list. I would add a comment type (print: string, var, var). That would log the string (perhaps a printf style) to a file with the parameters (var). I've always like printf style debugging.
lerman yea... I was thinking the same... even an extention to the msg command
lerman: would allow single step type debugging
sort of. step goes one motion, not one line of gcode
(and the interpreter ran a long time ago by the time you're stepping)
so when it hits a message line, has it actually done any moves that are past the line requireing a confirmation?
I don't know if message is a queue stopper or not, I think it puts up the message and then zip right by
hmmm could be.... I think is some of the gui's it doesnt even pause execution, just displays the messsage
axis needs a click on ok, but I think mini just displays the msg
I doubt any of them pause execution
03jepler 07HEAD * 10emc2/src/Makefile.modinc.in: fixes for external building of modules in simulator
my mistake... looks like screen update is paused until ok is clicked (or dialog box is moved forcing a repaint)
plus some extra time
see you Jymmm
or buy alex for $1.95 =)
* alex_joni gets two more
and a spare for parts
oh gawd... hell no!
we have enough alex's in this world for 10 lifetimes! lol
* alex_joni doesn't
and by alex's, I mena YOU!
could use a clone or two
couldn't we all
but do they have to eat and sleep too?
killem.. get new ones
and can you write them off on taxes and dependants?
Hey, you guys have holes poked in your firewalls... do you use JUST ssh or something else too?
yep that's about it
well, I run a webserver and smtp server too, but that's how I log in to my home system, and that's how I do stuff like IRC and read e-mail when I'm not at home
Hmmm, I wanted to toss a box into the DMZ, but I fear that if it ever got rooteed, it could be used to attak the boxes on the LAN.
Jymmm: use port knocking
no holes open
alex_joni I dont have anything open right now.
you use a program to access an sequence of "random" ports
ah, never heard of it
the prot knocking server checks the firewall logs
when it sees the right sequence it opens a port
for a short while
'security by obscurity'
when in doubt (and when not) use real security instead
Jymmm: there are various solutions out there
cradek: this is not instead
this is besides
I'd still open only ssh
but have it closed by default
most of this ssh advice is good: http://thinkhole.org/wp/2006/10/30/five-steps-to-a-more-secure-ssh/
but better than "disable root logins" is to use "AllowUsers" and list the only people who should be able to ssh in
advice #4 is terrible
and #5 is dumb
DenyHosts 2.x adds the very interesting (and completely optional) ?synchronization? feature. You can choose to send your list of denied hosts to the DenyHosts servers. They will, in turn, pass their collective list of blacklisted hosts to you.
sounds like sshcop
alex_joni: sounds like denial of service
sounds like distributed idiocy
SWPadnos: collective idiocy
no, it's central
collective peer-distributed idiocy
peer-collected centrally served idiocy
maybe one should add the DenyHosts to the blacklist
and send it to them
wonder what happens?
they wouldn't be able to log into their own servers?
SWPadnos: that would be too nice
only that and it would be worth installing that crap
at least there's no such crap in the dapper repositories
people will do really stupid things to avoid a couple hundred lines in a log file
or to avoid choosing strong passwords
hey - political question for those other Americans here: how many political parties do you have in the various races tomorrow (for Governor, house, senate ...)?
real ones or pretend ones?
well, real ones, including those who have no chance of getting elected, but not necessarily those who are just plain wackos
I don't understand the question :-)
in some races we have four: r, d, green, nebraska
I was watching debates last night, and there were 8 candidates for a single senate seat, 6 for the house and for governor, but only 4 for lieutenant governor
nebraska is wacko fundamentalists
green is ... well you know
the other side ;)
Here in Connecticut we have (at least) three Republican, Democratic, Independent (for Senator). The R is running third in a two man race. Most likely Lieberman will win (as the Independent).
I liked the "Impeach Bush" party and the "Anti-Bushist" party :)
then there's the crazy lady who's running for both governor and senate, with her main campaign issue being the legalization of Marijuana
I didn't see it but a coworker reported about a televised debate where the nebraska party guy only answered "I dunno but I sure do love Jesus"
sadly, that'll probably get him a lot of votes
to be fair, it may have been a paraphrase
they do get a few, but less than green I think.
more like "the love of Jesus will tell us teh answer"
yeah - I was sad that the Vermont Green candidate was actually the best
but a vote for him splits the non=republican votes, and gives mega-dickhead a better chance
well you can vote for him, then the democrat in the runoff, right...?
yeah, if we lived in Iraq or something
at least they get to execute their dictator over there
anyway - sorry to interrupt the ssh discussion with mundane stuff - gotta go to a meeting about reconstructing our street :)
You meant war criminal; not dictator.
alex_joni that port knocking doesn't sound too shabby, sorta like a "passcode"
I wonder if there's a OTP(ort) knocking ? lol
alex_joni do you use it? or jsut heard of it?
* alex_joni planned to use it someday
I read about it and decided I like the idea
[22:46:18] <Jymmm> http://www.portknocking.org/
googled a bit, found some clients
I like it too
I bet there might be an iptables + pam thing
[22:48:27] <alex_joni> http://www.portknocking.org/view/implementations
If the port knocking itself is setup like OTP sorta thing, and you had to knock (lets say) six times, 65000^6 that would be: 75418890625000000000000000000
you can increase that by using rotating rules
And eliminate (lockout) the connections that attempt to portscan or direct access to port 22
SWPadnos_ is now known as SWPadnos
You could even setup a double pattern... 132,132 to prevent the random portscan's
I like it!
I suppose you could enter timing constraints too :)
p1.. wait 1..5 seconds, p2 .. etc
Not sure about that, ip/tcp aint the best thing in regards to timing; dropped packets and all
yeah... probably so :)
but you could have certain patterns and repeat of patterns
and ones that change with time as well, but you would have to have a sync'ed clock
I did come across a soft token, but sounds if'y at best
maybe change the pattern on every successfull knock
even increasing one of the port numbers should be enough :D
* alex_joni goes to bed
nigth Jymmm .. don't knock yourself out
This is a new install of 6.06 at smithy
Is there a wiki page that describes the process of creating documents for developers using make
you need to install some required packages (lyx-qt, latex2html, and possibly others) and configure --enable-build-documentation
not sure what else you're looking for
I was thinking I'd create that ability on this box and document the process.
debian/control.in lists these packages that I think are the required ones for building the documentation: lyx-qt,python,tetex-bin,latex2html
I don't think there's really a wiki page on preparing a system to build the documentation, or how to contribute to the documentation -- that would be a great addition.
Okay I'll see if I can start one.
after changes are checked in, within a few hours the HTML and PDF format documentation is rebuilt on the linuxcnc website: http://www.linuxcnc.org/docs/devel/
when we release 2.1, the version on the 2.1 branch will get the same treatment, at http://www.linuxcnc.org/docs/
Okay Thanks jeff.
03jepler 07HEAD * 10emc2/src/hal/classicladder/projects_examples/ (example_many_sections.clp example_sequential.clp): *** files added for new_cl_branch ***
03jepler 07HEAD * 10emc2/src/hal/classicladder/ (7 files): *** files added for new_cl_branch ***
03jepler 07new_cl_branch * 10emc2/src/Makefile: port of classicladder 0.7.100 to emc2 hal
03jepler 07new_cl_branch * 10emc2/src/hal/classicladder/projects_examples/example.clp: port of classicladder 0.7.100 to emc2 hal