I still don't know what goes on the flip side
how much massaging are we gonna have to do to jepler's page to make a half-sheet double sided pdf with our URL and a few other details added
will the reference fit on one side?
jmkasunich: not much - he did it once already
that's a darned good question which I don't have the answer to
for the back, I have a few suggestions
drawings showing the definitions of G2 and G3 arcs
diagrams for G2/G3 was one
drawings showing how tool offset works
is the axis quick ref already on the G-code cheat sheet?
I can do drawings in easycad, but export to anything other than a pdf is tricky - we want drawing objects that can be placed within a larger pdf
I can do postscript from autocad
who has pdf authoring tools - to merge text and drawings?
I suppose lyx could do it (shudder)
I could maybe do it in LaTeX assuming we have postscript to start with
but it has a way of giving crap pdf output sometimes
don't know if that's changed in the last, uh, decade
[00:06:38] <jmkasunich> http://www.scribus.net/
ubuntu package exists
does pdf output
also, open-office wordprocessor might be up to the task
yep - one-click pdf export that actually works well
it's the image imports that could get screwed up
I always forget about open office for some reason
I can give Janet a call tomorrow to see the lead time.
perhaps it's the stigma of big bloated software pavkages
well there won't be any images, right?
I mean they'll be some kind of vector thing
unless the G2/G3 diagrams get in there
vector images, right
well let's try
I have to switch machines, brb
if there is a left over 1" square, might be fun to stick chips in there
these are half sheets, so about 5x8?
5.5 x 8.5
printable area is less though I suppose
the lamination needs to overhand the paper, I dunno if they are giving us half sheets of paper, and a little more laminate, or half sheet of laminate and a little less paper
they can likely print to the edge
but we don't want that, I imagine
5x8 gives a 1/4" all the way around
half sheets printed, with oversized laminate afaik
which seems fine to me - we're paying for the square inches, might as well use them
I know they won't offset print these - it'll just be laser printed or similar
that may mean that we don't have a full bleed
yeah, there will be borders
question is how big
1/4" seems reasonable - most modern printers can get pretty close to the edges
I can ask. they may actually be able to do full bleed though - I don't know
well, keeping a smallish border makes alignment less critical
this is black on white, right?
so for the drawings, we can use line width, and _maybe_ shades of gray, to distinguish between things like part shape and toolpath, etc
gray will work, it'll just be dithered in the printer
did anyone use the Altera FPGA toolset for linux? ( Quartus II web edition)
tomp2: I used the windows version -- I was under the impression that only paid versions were available for linx.
[00:47:05] <jepler> https://www.altera.com/support/software/download/altera_design/quartus_we/dnl-quartus_we.jsp
says "For Solaris or Linux support, purchase an Altera software subscription"
app = Tkinter.Tk()
app.title('FPGA Configuration Editor')
resulting window is 305 x 267
[00:52:18] <cradek> http://timeguy.com/cradek-files/emc/arcs.eps
the outside box is 5x8"
jepler: thanks, i saw that 'buy a subscription' thing too... :(
R arcs blow up if R is less than (or too close too) 1/2 the dist from start to end
it's ugly, but I can't put a finger on why
also, I didn't make sure it's right, someone else should check it
IJ arcs blow up if the IJ isn't equidistant from both start and end with some tolerance
thats how I read it, as someone who has no clue how arcs work
did I get it right?
also R arcs are a bad idea if the endpoints are close together
then it must be a good drawing ;-)
close relative to R?
nearly a full circle and nearly a half circle are the bad cases
they get a bit unpredictable
i'll try scribus if you'd like
tomp2: if open-office will do the trick, we'd probalby prefer that
ok, will look at oo and cradek's eps
cradek: regarding the ugly... true, but I think thats the tradeoff for information density
you got 6 possible arcs in there
oo impress or wordprocessor?
I'm displaying it at 3" x 1.5", and its quite readable on screen - paper should be better
I got a pdf with eps2pdf - is that why we are trying OO?
can you merge text and multiple drawings and get the layout you want?
oo gives you wysiwyg, fonts, etc
I could draw more stuff in autocad, but its fonts are ugly
I doubt I can merge/import anything
we certainly don't want the entire g-code reference in acad fonts
oo will do that nicely in a table I bet
the gcode reference is done, jepler generated perfectly good postscript
no need to rework that
already scaled to fit a half-sheet?
we've only got to worry about the back side
yes he showed me a printout, and it was easily readable
jmkasunich: can easycad do better fonts? I could give you a dxf to fix up
cad fonts are never nice
typical easycad fonts: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/~checkout~/emc2/docs/src/hal/encoder-block-diag.eps?rev=1.1;content-type=application%2Fpostscript
maybe they are nicer
is your plan to do the entire thing as one drawing, 5x8"?
I won't be offended if you want to redraw it
I had no plan, except I thought I had a good idea for how to present the arc information
ok, lemme put it differently ;-)
how do we want to do this? one 5x8" drawing, or several EPS snippets that we lay out on a page?
I still don't have an answer to that :-/
I think I don't have/know the tools to do this
if the back is gonna be 50% text and 50% drawing, I don't think we want it to be a drawing - writing paragraphs in a CAD tool sucks
well consider that file my suggestion as to how to present the arc information we want
can you do something similar for tool comp? and lathe tool offsets/shapes?
(I think there is something for the latter on the wiki)
I'm exploring layout possiblities
scribus docs author is opinionated: "No matter which way you spell it, a Tagged Image File Format is the file format for bitmap images. Period. End of story. Don't give me any arguments: I'll win."
that's one word for it
jmkasunich: I added "import Tkinter" and "app.mainloop()" to your python above, and I got a window that was around 800x600 pixels
* jepler disappears again
hmm, then I'm obviously doing some thing wrong elsewhere
are you putting widgets into 'app' with 'pack' or 'grid'? if 'pack', then you need: app.pack_propagate(0)
that's the only gotcha I can think of
bet thats it
work_area = bwidget.ScrolledWindow(app, auto=Tkinter.BOTH);
you used grid in your example of making 20 widgets in a loop
I haven't gotten that far, I will probably want grid when placing stuff in my work area, but maybe pack for menus and placing the work area in the main window?
not sure, I'm really out of my depth
changing app.grid.propogate to app.pack.... made it big
but may not be what I want later
cradek: when you exported arcs.eps, did you specify the page size somehow?
the image is 8x5, even tho you only drew on about 2.5x5 of it
I specified 1=1 scale
[01:40:16] <cradek> http://timeguy.com/cradek-files/emc/arcs2.eps
is this one wrong the same way?
you're using more of the paper, thats good
I think it might not matter
* jmkasunich is scrabbling up the learning curve here
[01:43:13] <cradek> http://timeguy.com/cradek-files/emc/arcs3.eps
are you using line weighs or width or something?
maybe this one is right?
I told it 5x8 paper size
yes I used some thicknesses
they do show up in evince...
(I've never used acad's ps out before)
does dia do svg output?
does dia suck to use?
there's always xfig but it sucks like all the other programs
I think its a simple drawing tool
I'm trying to think of what other pictures a gcode user would want to see all the time
the lathe tool stuff
aren't there three parameters and 8 orientations or some such?
I don't know if that's in the same class as this stuff - you have to look at it when you set up the tool table, but it's not a gcode issue
a g76 diagram would be great, but I'm not sure I'm up to it
I'm not even sure a picture helps with that. Maybe the text explanation is better, since a picture would get so busy
btw, is jepler's PS for the front online somewhere?
I bet not
* jmkasunich pokes jepler
I think he just printed it from firefox, and maybe massaged the output a bit
I can't figure out how to get it onto one page
get what. the g-code ref?
[01:58:30] <cradek> http://timeguy.com/cradek-files/emc/gcode.ps
wonder if I can get rid of the stupid headers
yes, grab a new copy
hmm, I notice that the coverage of O words is very limited
probably not something that can be done justice to in 4x5"?
the basic syntax could be covered I bet
1st run a OO's Impress http://imagebin.org/8833
tomp2: I think cad does a nicer job, and can be scaled easier to fit
i agree, dtp sux for cad drawings, while cad sucks for presentation, go figger ... cad is better suited for this task
btw, that oo was fully scalable ( tho jaggy )
cradek: when you wrote "G2+R" that means G2 with positive R ?
vmware is frustrating
one of the nice things about easycad is that it is incredibly responsive
in vmware, its not
I'm forever clicking on the wrong thing because the mouse cursor is a fraction of a second behind
I was having mouse trouble too tonight
I think I'm going to try vmware with it connected to a real serial port, and use a mouse or tablet directly on that
jmkasunich: "G2 R-" etc. might be better
tomp2's penguin is cute though
yeah, since in the gcode you'd say something like R-0.1
you are sure that your drawing is correct? positive R = take the shorter path, negative R take the longer one
to be clear: the four arcs would be G2 R+, G2 R-, G3 R+, G3 R-
you expressed some doubts earlier (or I mis-read)
I was sure, but I also checked it
I was hoping someone else would check it, but nobody did
(because there could still be a thinko in it that I'm blind to)
the I/J assume the XY plane, right?
would it make sense to write (X, Y) instead of (or in addition to) END?
yes (to both)
that way all letters in the command are shown on the dwg
for a straightline entry move, you'd normally make the move toward an outside corner of your part, right?
I don't know what's normal, but often you come in at a corner, tangent to the edge
I think this is as distilled as O words get
the line numbers are used to match up while and endwhile?
make them different in each example
not the first 2
oh, not for looping
what is the syntax for a condition?
[03:24:55] <cradek> http://pastebin.ca/535572
I'm the perfect person to examine this, cause I don't know it
the call sort of looks like it has positional parameters, is that true?
if you provide 2 params, the over-rule the normal values of #1 and #2 for the duration of the subroutine?
conditionals are lt, gt, le, ge, eq, ne
the call does have parameters, and they do scope like you say
and the operands for those are #vars or constants, right?
well they are expressions
#34 lt 12.5
like , [2/3], [sqrt], etc
oh, the  are part of the syntax, not part of your document notation
[#34 lt 4.5]
o100 while [#34 lt 4.5]
what you've written will fit in 4" tall by 2" wide
could you write something about expression syntax in another 4x2 box?
* jmkasunich buries cradek with work
I have to be at work early too
do it at work tomorrow
I will be up another couple hours, it would be nice if I could finish the page
I don't know yet how to merge the text and drawings
worst case, I do the text in cad
[03:34:52] <jmkasunich> http://jmkasunich.dyndns.org/pics/G42.png
does that look right?
solid is programmed move, dashed is actual, hash lines indicate part outline
I'm really tempted to try to show entry and exit moves, including arcs for offsetting inside a concave pocket
except I have I hunch I would make a nice pretty drawing that is wrong
atan takes two operands? or is it acting on the result of the divide, with no outer set of  ?
jmkasunich: yes your entry is right, except maybe it should be a bit longer - needs to be more than a radius
atan is a special format, sort of takes two operands
it's actually like atan2(y,x)
but you actually write the / ?
it's the only "special" one
yes you have to write it just like it shows there - it's an error otherwise
can you write SIN 30, or is it SIN?
so when do you not use 
X3 doesn't of course
you use  to get expression evaluation
and to mark off arguments to things
IOW, the only time you don't need  is when you have a single constant, and its being passed to a word like X, Y, etc
yeah I think so
[#40 GT 12.5] or [[#40] GT 12.5] or [[#40] GT [12.5]] ?
any of them
I would NOT have guessed that
even indirect: #[#40]
don't mistake me for an expert - if in doubt read the spec :-)
we should have started this a week ago
hey, I know, lets do the cards next year
nope, I want this card now
I'd use it all the time
damn, I wish I knew what I was doing
I'm drawing a square part with a square pocket (rounded corners)
I want to show entry and exit moves inside and out
inside will be arcs, since there are no convex corners
outside entry move is easy, you can come into one of the corners in a straight line like you drew
inside entry move will have to be an arc, radius bigger than the tool
right, and exit looks the same
exit is nothing special - just move away from the part and turn comp off after you're away from it
I wouldn't even bother to show exits
how does the tool return to the nominal path?
in the move after you turn off comp?
the move after g40 puts it back on the path
I want to show the moves, so I can mark the spots where you invoke G4x
for exit then, show a move past the part, then g40, then another (straight) move over which the tool returns to the nominal path
and inside I can straight or arc away, then G40, then straight
in real life you'd probably move up and not screw with exiting inside the part
you'll have to arc away, since you can't make a corner
for the entry
(that's why it's easier to move up, since you can exit with a straight line later)
the programmed move will be an arc greater than the tool radius
the actual move will be a tighter arc
ok, got it
that case is the one I drew
I don't mind cadding till 2am if needed, but cadding till 2 am and having someone tell me I borked it up tommorrow would suck
yeah I definitely understand
you drew a straight entry move didn't you?
that's what you should do I think
why? don't arc entry moves work now?
the arc is what gets to the part, and is fully compensated
yes they do, but I hear no other controllers do them :-)
this isn't for other controllers
up to you
I think our single arc entry is elegant
so do I
so I want to show it
just don't show an exit with an arc - not so elegant
exit will arc away, then G40, the line away
remember to move off the part before doing the g40, and after the g40, make a straight move with len > radius
you're going to be a gcode programmer whether you want to or not :-)
if I ever get my machine built, I will
anything else before I sleep?
I'm sure I'll think of something in an hour or so
I can proof or help more in the morning
SWPadnos: can you find out what the printer needs? is one pdf for the front and a separate one for the back OK?
SWPadnos: obviously thats an answer you can't get till tomorrow - but email it to cradek and I as soon as you can
Got a question on offsets and the var file.
Both tkemc and mini have an offset display where the user can enter values.
I'll try to answer your question, but be warned that my office's internet connection is going down for maintainance any minute now
so if I go silent that's why
When those values are made current, tcl makes a copy of the xx.var file, fixes the new values, writes the file to disk, and issues an emc_task_plan_init to bring the interpreter's variables into line.
Is there a way to turn the process around so that a G10 L2 (num) x y z is issued to fix interp vars and then write from the interp to the file?
Oh. I found Interp::save_parameters. Perhaps I can use this?
hi -- connection is back for the moment
in AXIS we do use G10 L2 to modify coordinate systems
I don't think we do anything special after mdi'ing the G10 L2
though maybe that's hidden inside the call to "reload_file" which recreates the preview according to the new coordinate system..
clear_command = "G10 L2 P%c X0 Y0 Z0 A0 B0 C0\n" % num
^^^ for instance, here's how we clear a coordinate system to get rid of any offset
hmm -- *something* we do makes the .var file on disk be updated, because that's how the offset is applied for the preview plot
there are a number of mode changes and other calls here in axis .. I'm not sure which one is responsible for causing the .var file on disk to be updated.
Well at least we know it's being done.
is it important to tkemc to be able to read the .var file and see the right offsets?
Only if something like abort causes a read of the file.
if not, maybe you don't need to solve that portion of the problem
That seems to be where g92 gets some of it's pecularities.
You are right that the tk stuff does not change anything graphical based on changes to offsets.
it looks like the interpreter writes the var file in response to EMC_TASK_PLAN_SYNCH_TYPE
and also in emcTaskPlanExecute when it's an MDI command and "in position"
so basically I think the .var file is written as soon as you MDI a G10 L2, assuming the machine is not moving at the time, due to an earlier MDI command
Ah. That'll do it.
rayh: glad I could help
Guest947 is now known as skunkworks_
oh no, there's a bit of drama in Galesburg: http://www.metafilter.com/61766/You-Cheer-You-Lose-Your-Diploma
impotent authoritarians everywhere
CHEERS COULD HAVE COVERED THE APPROACH OF A TERROR ATTACK DON'T YOU PEOPLE KNOW THERE IS A WAR ON AND THEY ALMOST TOOK OUT JFK AIRPORT LAST WEEK 9/11 CHANGED EVERYTHING WHY DO YOU HATE AMERICA SO MUCH?
indeed I laughed at that response too
remember - cheer for your enemies, and they won't get diplomas!
SWPadnos: are you going to bring an AMD64 machine to fest that we can pork up with a realtime kernel?
jepler, I'll have my dual Opteron amchine
I don't think I'll be bringing a single-core unit though
the UP realtime kernel ran on my 2-core machine
ok. we can try that, and also try to get one based on cradek's SMP build to work
what OS(es) on that machine?
seems like you said something about 64-bit feisty the other day
I can bring some hard disks
cvs merge --smart cradek-smp + jepler-64bit :)
that one has 6.06
I have another with feisty, which is meant to be a fileserver, so I may not bring it
that's a single-core Athlon64
I wonder when the next LTS will be -- 8.xx or 9.xx?
every 3 years, I think
so around 9.06 you think
only if there's a dealy
I wonder if our "LTS only" position for emc will remain tenable until then
the releases are supposed to be every April and October, but they were late last year
note also that the "desktop support" is only 3 years - 5 years is for server only
dV/dT = I/C ?
too bad I forgot even the parts of my differential equations class I attended
differential equations: guess something with an e in it, check, repeat
yeah, but how do you check your answer?
I don't remember
it's strange that I remember a lot of trig, a lot of calculus, little statistics, and no DE
I don't know the titles of the few things I still remember
it seems related to how well I liked the class/teachers which is disturbing
and conversely, grades are dependent on how much the teacher likes you ...
all I can say is yay for the printing press
yay for google calculator
does google calculator solve 'dv/dT = V/RC' ?
I don't think so
jepler: V = e^-kt, discuss
I tried something with sqrt() in it, and it gave me a normal search
I think that's e^-3/2kt
or was that something else?
k ~ 3/2k
(for large values of k)
k ~ RC
so R=3, C=1/2
1==2, for some values of 1 (and 2)
if you need anymore help, let us know
emc needs some kind of generic "transfer function" component(s)
a bit more seriously, hal_polynomial order=3,4,...
loadrt hal_polynomial order=3
heh, that must be obvious
setp polynomial.0.a ...
anyone else think we should quit with the hal_* prefix?
for a while I thought I wanted something like that for my spindle pwm, but it turned out linear enough
I can see that it might be nice for some (obscure) things
I'd like to make a 2.1.6 release and new CD this week
any thoughts about the state of the 2.1 branch?
I agree, i'd like you to do that
I haven't deliberately introduced any bugs
I think the motion carries
I'd also like it to be the last 2.1 release (but I always say that)
do you suppose the features in TRUNK are good for 2.2, and we should discuss 2.3 things next week?
if only I could remember what was in TRUNK
I think I added some things
all that extra stuff
I fixed some things
I think I even added and/or fixed some things
it must be very special now :)
I'm not using 2.1 for either machine - it's pretty useless for the lathe actually (tool table problems)
anyone know if there's stuff in the sf tracker that is serious and affects 2.1.5?
argh why can I never get what I want from 'cvs log'?
huh is this all we did since 2.1.5? http://emergent.unpy.net/files/sandbox/emc-2.1.6-changes.txt
(generated by 'cvsps -b v2_1_branch -r RELEASE_2_1_5 --root :ext:cvs.linuxcnc.org:/cvs' emc2 since I can't get cvs log to do what I want)
yes I bet that's it
there isn't much
we're pretty sure the stuff for elson's boards is right now?
I asked Jon E to test ppmc index and make sure the sample config is updated
I decline to answer that except to say I asked him to let us know
well those changes seem to cover the items listed on the wiki as "known problems with this release"
Known problems in this release:
* The ppmc index pulse polarity is reversed
* tkemc keyboard shortcuts for incremental/continuous don't work right
it's looking like the end of the line
but in a good way, right?
Merging differences between 184.108.40.206 and 220.127.116.11 into changelog
rcsmerge: warning: conflicts during merge
hm I must have something in my local 2.1 tree
I wonder what it could be
for some reason, emctask.cc refuses to accept subsequent EMC_TASK_PLAN_OPEN packets after the first one: "emc/task/emctask.cc 270: interp_error: A file is already open". I already called EMC_TASK_PLAN_CLOSE, so is this a bug, or am I missing something?
I turned on debugging information while running tkemc.
I observe that it sends the following messages when I "open" a file:
Issuing EMC_TASK_SET_MODE -- (+504,+16, +16, +2,)
Issuing EMC_TASK_ABORT -- (+503,+12, +17,)
Issuing EMC_TASK_PLAN_OPEN -- (+506,+268, +18,/usr/local/jepler/src/emc2/nc_files/arcspiral.ngc,)
Issuing EMC_TASK_PLAN_SYNCH -- (+516,+12, +0,)
jepler: I did the same for keystick, too, and I saw that the keystick code only calls task_plan_open (and nothing else)
hm, that synch message looks interesting
It's not clear to me that EMC_TASK_PLAN_CLOSE actually does anything
that might explain why it's not showing up in my debug output
a little interactive session in Python with tkemc running shows that the ABORT may be what enables the subsequent open
emc/task/emctask.cc 275: interp_error: A file is already open
hm, didn't notice that before. thanks.
the full Python session, not cleaned up for shorter paste into IRC: http://pastebin.ca/537744
though the differences between the Python interface and the C++ interface are large (particularly with reslect to naming), a quick interactive Python session may be helpful in exploring what calls are needed to accomplish something -- without the need to re-link a C++ program
good night all