SWPadnos_ is now known as SWPadnos
* alex_joni hugs CIA-20
cradek_ is now known as cradek
* alex_joni kicks CIA-20
SWPadnos: feel free to attempt to try to implement that
so you didn't have to stop the motion queue unless the interp got to someplace where it needs the value, and sees that it doesn't have it yet :)
har har :)
SWPadnos: next step would be branch predictions
d need a var for each axis and then MDI is history.
and other CPU-like improvements
how about simultaneous execution units?
like XYZ and UVW?
I always wanted something like x2*#y
ha.. link axis.1.pos-fb to motion.analog-input-01
and presto :D
(but without offsets & all that g-code crap)
hey ray, and probably jmkasunich, I may need to ask you guys some questions about the prehistory and history of EMC on Linux to prepare for my ESC talk
in the next month or so
sure... my history only goes back to 2003 or so.
you can also ask JonE .. I think he also has some history
then again, you wrote aRTAPI, right? :)
yep that's true
* rayh has old-timers and a rose colored history filter
fred wrote the original, I modified it
I should email him too, once I figure out what I want to ask/say :)
maybe ask him what to say?
oh, good idea
rayh: after four years of "working" on my CNC conversion, it moved under computer control for the first time the other day
jmkasunich: forgot to congrat you on that..
saw it on your blog today..
so far lathe only, haven't mounted or wired the Z motor yet
heh - that's almost inspiring enough for me to brave the cold in the garage to work on mine :)
That is a great feeling when it happens.
brrrr - nothing could be that inspiring
it's up into the 20s I think
At -15f it takes a bit of courage.
oh my god - it's freezing here!
today's mission is to get the spindle motor contactors hooked up
Had to go out to the shed and pick up some acme. It turned white when I got it in the basememt.
that is the "don't touch me if you want to keep your skin" warning
Is this mesa connected, John?
parport so far :)
rayh: not yet
It was calling out, "lick me quick."
a surplus db-25 to 25 screws thing with no circuitry, plus a perfboard that I made yesterday with two relays, three optos, and a charge pump
haven't wired in the perf yet
rayh, hmmm - that brings up an interesting question - do you have a "favorite" parport breakout?
or least favorite, as the case may be :)
I've talked to a couple of people lately who were looking for breakouts, and there always seems to be some deficiency
from pictures I like the pdmx one that sits on geckos
that's great for steppers :)
That's what the solder fumes comments were about. Good on you, John. That's how I did my first ones.
I used the gecko pmdx on Smithy's new desktop mills.
I like it.
seems like that one saves quite a bit of wiring
If you use gecko drives.
and don't care about gecko's shifting definition of fault.
the older drives still have the same fault output though, right> it's just the 203V that has a different function?
No some 2xx had a fault output some didn't.
hmmm. maybe the G202/G212 had a new scheme
bbl.. really need to crash now
see you alex
I'm the eternal optimist - I've already started collecting parts for my next project
[16:52:13] <jmkasunich> http://jmkasunich.com/pics/newparts.jpg
passive with a pci parport card
(only 82k pic for those with slow wires)
jmkasunich: what is the servo going to be used for? spindle?
That'll take 5 minutes!
its 1kw continuous at 5000 rpm
the table is 12 x 24" and 125 lbs
I'm hoping to make a bridge mill of sorts, and have it come apart
so the biggest chunk will be base/table, at maybe 150-175
so I can bring it to the CNC workshop
not in 2008 tho
bridge or gantry?
whats the diff?
I'm thinking stationary table and moving bridge
bridge -- x slides under
that would take almost twice the bench space, so I was thinking the other way
gantry is a stationary table.
gantry then, although nothing is cast in stone
That way you just add the width of the gantry to the length of x
How much Z you thinking?
I saw a mill on the net that did moving table, but they moved it the short way
I'd like about 10"
a vote - stationary table with slides mounted to the long direction.
skunkworks: thats the current plan
might put them right on the side of the table
medium sized thompson linear rail.
hanging over a couple inches on each end.
I have 8 THK SR20 slides (no rail) already
I think its sr20, maybe hr20?
and about 5" between slides
I wonder if I need two screws for a span that short (probably about 14" with the rails)
I've been assuming I do
I most always err on the side of too heavy.
I would.. either 2 motors or a nice timing belt.
if I didn't have the desire to take the machine to CNC workshop, I'd be erring on the side of way to heavy
for the gantry, I've been toying with the idea of doing it like a planer - the crossarm moves up and down
that way when you are cutting 3" above the table, Z isn't sticking down 10" from a bridge that is up 13"
of course that would mean two screws for Z as well
that'll need two screws, I think
A guy named tony used to hang around here.
He and I built a bridge router that way.
Worked a treat.
stiffer than traditional bridge routers?
Single motor and timing belts.
the part that currently has me stumped is B - I want to be able to tilt the spindle, so it can be a vertical, horiz, or anywhere in between
I like horizontal for work that allows it - very nice chip clearing
are you going to make a setup just like max?
actually, I think I'd try to do AC, not BC
wait, thats wrong
I want BC
but my support is to the left (or right) of the spindle, not behind it like max
AC should allow the same geometry as BC, shouldn't it?
yes you need any two rotaries
except you rotate the support rail
(in this configuration anyway)
you just have to mount the work differently
A can be vertical rotary on the table
or in this case, "nod" of the head
this gets confusing because even X and Y aren't well defined yet
traditionally X is the long axis
that's a big motor though, a table-mounted rotary is probably easier
but for a bridge machine, you typically stand at one end, not the side of the table
SWPadnos: good point about the motor - I'm not committed to it yet
it seems big for a small machine :)
I'd REALLY like to be able to tilt the spindle
I already had it, so it is my current leading candidate
yeah, similar to the bridgeport head adjustments
but I'm keeping my eyes open
I don't really need a kilowatt I don't think
AC trunion seems easy to build
you need plenty of Z travel though
cradek: I'd need another rotary
ebay often has AC servomotors/drives cheap - like $100-200, in the 400-watt range
and that still means vertical spindle
I like the idea of a horizontal spindle so the chips clear nicely
the Yaskawa drives can do position or velocoty control, so they could be used for a spindle
thats why I was asking you about them last night
then you mentioned the price
the ones I have are 50 and 200 watt, so a bit small
so far, I have $50 in the table, $40 in the slides, $150 in the rotary (bought that years ago), and $0 in the motor
I don't see why AC trunion means you need a vertical spindle
I know its gonna cost a lot more before I'm done - $hundreds in screws and linear rails
with a horizontal spindle it would be very nice to rotate A90 and mount the part sitting flat
cradek: if I want to do a largish flat part (3axis), I want to be able to remove the rotary and attach it to the table
(like a 10 x 20 panel or something)
9 minutes to go: http://cgi.ebay.com/YASKAWA-SGDH-04AE-SERVOPACK-LOT-OF-2_W0QQitemZ280183451139QQihZ018QQcategoryZ78191QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
you 'only' need a big angle plate
and matching Z travel
the angle plate would be the heaviest single part then
hmmm. maybe I should buy those - they're useful for the disc motors as well
I wonder how much spindle power I really "need"
well, that depends ... ;)
no matter what you have, theres always a part or cut that wants more
but for real heavy stuff, I have the van norman with the 5-1/2" face mill
and then there's all that titanium you want to mill
so I don't thing this machine needs to swing anything large
I am interested in small tools - I want to go all the way up to PC board milling stuff, which will require a very fast spindle
I'm thinking belt driven from the main spindle
I think I have a visitor, back soon
that was my carpenter/magician friend, picking up his most recent gimmick
Sworn to secrecy - right? ;)
he uses tools for props
he gave me a 3/8 fender washer
and I made a new one that is a tiny bit bigger and a tiny bit thicker, that the real one nests inside of
about 0.010 wall thickness
nice. that is a pain to hold.. did you make the reccess/hole - then cut the outside diameter
no, did the OD first, then stuck it in a soft collet that I bored to fit
fortunately the new part is just a tiny bit bigger than the last thing I used that soft collet for
* alex_joni hugs CIA-20
* CIA-20 hugs alex_joni
if only we would all return hugs but not kicks
* cradek kicks CIA-20
poor bot :)
* alex_joni rubs CIA-20's tummy
ok that's getting a little strange
* alex_joni eats CIA-20
* CIA-20 tastes crunchy
* alex_joni stops now
you can still 'kill' it :)
jmkasunich: what people do for fun :)
ancient remains from when cia was a smaller project ;)
so those responses aren't ales hijacking cia? they're built into the bot?
jmkasunich: you think I would do that?
* jmkasunich hits CIA-20
* jmkasunich kicks CIA-20
great, that means I'll soon have to disable those things because everyone keeps messing around with it ;)
BearPerson: you are a member of the CIA project?
depends on how you look at it, but I guess it's mostly a yes by now
cool - its a really nice tool
I hope I got rid of most of the basic bot trouble
* alex_joni votes for yes :)
I've been lying to CIA for a long time now
how so? :)
I'm running the emc compile farm, and when a build fails, I send a message to CIA so it pops up here
its not really a commit, but I fake it with the content that I want to appear on channel
hmm, interesting approach
I should turn that into a feature sometime ;)
[20:36:13] <jmkasunich> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/infrastructure/farm-scripts/results_cia?rev=1.3
that is what I'm using today
if CIA wanted to support an "announce" message (or something along those lines) in addition to the "commit" messages it supports now, I'd change my scripts to use it
jmkasunich: the one good thing I can say about your script is that it works
the only good thing?
what is IMAGE, ALTTEXT, etc?
where are you seeing that?
oh, those images are the check or X on the web view
I thought you were looking at results_cia
the webpage uses server side includes to stick the results into the main page
that is results_cia
results_cia has some remnants of that code lying around, though it obviously doesn't do much
warning warning copy-paste coding detected!
* jmkasunich fixes
EMC: 03jmkasunich 07TRUNK * 10infrastructure/farm-scripts/results_cia: removed some leftover code
looks like commits are working, too.
BearPerson: yup, for now it's all good
will bug you when it stops
EMC: 03jmkasunich 07TRUNK * 10infrastructure/farm-scripts/results_list: removed some more leftover code
good, thanks :)
jmkasunich: now if you could only make our naive cvs server not send the same cia message for each directory of a big commit
oh wait that's my abominati^Wbaby^Wfault
BearPerson: do many other cia users have that problem, or do you know?
"that problem" ?
(cvs is pretty bad at this)
every directory you touch with a cvs commit runs the "loginfo" program once. with a naive cia message generator, you can get several at once
indeed, that does explain some problem reports I got earlier
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2007-12-19.txt
not sure if I can find one quickly in our logs
I'm guessing it would be most elegantly solved inside the cvs commit hook, will have to see when I get around to looking at that
since we're on the topic, do you guys know which part of the process generates email subjects with tabs/carriage returns in them?
the 12/11 log should have a bunch from Francis Tisserant
clearly not cia
[21:02:55] <SWPadnos> http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2007-12-11.txt
look around 14:35
that's a good example
I wonder if esmtp wraps the subject (and maybe other headers)
my mailer doesn't care, it does the right thing. yours has problems with this?
it looks ugly in the message list :)
(wrapping headers is perfectly fine I think)
it's GUI mozilla, not a text terminal app
I uploaded some images a long time ago (when the CVS move was being done, I think)
mutt shows the split header in the message itself, but in the message list it unwraps it and shows as much as fits across the terminal
[21:08:22] <SWPadnos> http://www.cncgear.com/images/YuckyTabs2.jpg
those are from sourceforge...
same thing happens with CIA messages
I recommend getting a better mailreader
what could be better than Mozilla? :)
mailman, the mailing list software, wraps long header lines because some RFC says there is a maximum and recommends to keep them under a certain width.
see 2.2.2 and 2.2.3
I'd think that same RFC would say that carriage returns aren't valid, but that's probbly wishful thinking
at least in older versions, my mutt would show the wrap as "?" instead of a whitespace; it would also disturb the thread view because it though the various slightly-different wrappings were distinct subject lines..
ah ok. I should see if it's possible to look at enough of the Mozilla code to understand where to put the s/\r// expression :)
it will still have whitespace you may or may not want to keep
true, I guess something that collapses any whitespace group containing a CR into a single space would be better
Each line of characters ... SHOULD be no more than 78 characters
right, and MUST be no more than 998
which is pretty funny
yeah, considering that some computers hardly had enough RAM to hold a 1k line :)
well rfc2822 is from 2001...
the 78 character limit is really hilarious anyway
it's not like you're going to "print" an email onto a punched card
and Windows has the only console that can't do anything other than 80 (and possibly 40) columns
"to accomodate ... user interfaces"
any old laser printer will truncate those lines too ...
personally I think mozilla's freakout behavior is funnier than the rfc :-)
yeah, it is a bit strange
if they'd just print the wide characters in white instead of black, it would look more or less OK
looks like 'yahoo mail' and 'pegasus' are the only remaining programs used to post to our lists that are too stupid to get threading right
haven't heard that name in a while
last night I switched my mom from eudora to gmail. remains to be seen whether that was a good idea.
BearPerson: did you say that people report cvs's behavior of multiple messages as a cia bug?
yrah. it's hard to know if the web-centricity of gmail is a good or a bad thing
well, I remember people asking if I knew why cia sometimes "reported commits several times"
that sure could be it
not sure how proficient they were with cvs and therefore the exact source of the problem
many cvs loginfo implementations rely on timing hackery (all commits within so many seconds get grouped together) and that is obviously fragile
it's definitely cvs's problem but it would be nice if you could do something easy in cia anyway
not sure what we could do in cia that couldn't be done in a commit hook
I see it already has queueing and flood protection
well in our situation, cvs is run inside a very tight chroot for security reasons. The ability to do something complex just isn't there
the loginfo program sourceforge uses requires all of perl and temporary directories and files and who knows what else
but you are definitely right - it's not cia's fault - I'm not saying it is.
cvs is a dinosaur but a well-loved one still, it seems
once or twice a year we have a huge battle over switching to someone else. I hope I didn't just trigger one
or, something else, you might say :-/
can you tell what percentage of cia users are using cvs?
(but see caveat at left)
those bars are not to scale :)
no they're definitely not
unless it's an exponential scale
interesting that cvs is still the clear second
heh - bitkeeper, 1.84 yeaqrs ago
I guess nobody using git uses cia
no sccs? haha
git shows up as other
ah yes, I suppose it would
[21:53:16] <alex_joni> http://cia.vc/stats/vcs/other
-> check the frugalware commits from today
heh - different kind of too9ltip, I imagine :)
[21:54:43] <jepler> http://www.linux.com/?module=comments&func=display&cid=1137476
SWPadnos: you think? sexy-tooltip.c sexy-tooltip.h
oh man, I love your bull-nose 4-flute end mill
i wonder if git messages are underrepresented because they show up as change "sets" rather than, for example, a flood of cvs commits
ooo cia's search box is neato
yeah, it's quite nice