jepler has changed the topic to: EMC2 development -- http://linuxcnc.org/
| Latest release: EMC 2.3.4 | channel logged by logger_dev
darn, missed mshaver
had some fun tonight making some little parts - cut aluminum dry at high speed with coated carbide
I didn't have the spindle speed for that before
heh, I tried it that way because I don't have coolant nozzles yet...
when I tried that on the bp it just howled - this is cool.
also, having a tool changer is so nice. spot drill plus two more drills plus an end mill? no problem!
:) what is the BP going to be used for now? ;0
nothing until I reassemble it...
it has to be assembled to use or show it running for sale - it has to be disassembled to remove from the garage.
pretty sure I will reassemble it ... one of these days.
EMC: 03cmorley 07master * r89a89fef0915 10/src/emc/usr_intf/pncconf/ (pncconf.glade pncconf.py): Fix found tuning test problems
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2009-10-26.txt
cradek: archivist might like to test your patch..
I'm going to do some more sim testing... but if anyone wants to try it, I'd like the input.
cradek: is c63a4c862f50014d3512352fd7b164bb8ce2b98d suitable for 2.3?
I didn't look at the patch but the change message sounds like it might be
well it fixed the problem dewey found, and it didn't break any of the tests...
so it probably should be backported.
for consideration: add pin to clear axisui notifications: http://www.panix.com/~dgarrett/stuff/0001-create-pin-axisui.notifications-clear.patch
dgarr: I'll give it a shot
looks pretty straightforward
i just followed the pattern for set_manual_mode; thanks for looking at it
it's very useful to put up notifications and clear them automatically!
jepler: I remember you had a way to get an operator message sent from hal. did you do this by adding it to halui? I couldn't find anything in the archives.
cradek: use rtapi_print_msg(RTAPI_MSG_ERR). http://media.unpythonic.net/emergent-files/sandbox/error.comp
cradek: what do you want me to do with this touchy change of widget highlighting and mouse cursor?
[16:11:26] <jepler> http://emergent.unpy.net/files/sandbox/0002-don-t-show-a-mouse-cursor-pointer.patch
I tried the first one - it looks great
sure, I'll try it
(I tried unclutter, but it doesn't work well)
hm I've made a bit of a mess of my tree :-/
EMC: 03cradek 07v2_3_branch * r8dbbf87db900 10/src/emc/rs274ngc/ (interp_convert.cc interp_queue.cc): fix undetected gcode error when using cutter comp
Your branch and 'origin/master' have diverged,
and have 248 and 796 different commit(s) each, respectively.
ok, I doubt that's right
you done something wrong
yes, yes I have
EMC: 03jepler 07master * r8912b6c711c1 10/src/emc/usr_intf/touchy/touchy.py: Stop the pointed-at button from looking different
EMC: 03jepler 07master * r14f67c137c4b 10/src/emc/usr_intf/touchy/touchy.py: don't show a mouse cursor (pointer)
oops, I borken the 2.3 build
maybe I should have ... tried it
jepler: on what state defer-format branch is?
micges: I don't think I've touched it since the initial announcement
can someone refresh my memory: To build debs I cd to emc2/debian and...?
I want to test cradek's pATCH FOR THREADING
micges: I don't need defer-format for myself. I was hoping for feedback from others..
mshaver: something like debian/configure -a; dpkg-buildpackage -rfakeroot -uc -us
EMC: 03cradek 07v2_3_branch * r5ec0dbdaa6b8 10/src/emc/rs274ngc/interp_convert.cc: oops, fix build fail by silly macros
the wiki talks about something called pbuilder, but that's not how I remember doing it
mshaver: yeah, pbuilder is overkill
jepler: I know, I need defer-format for myself and I'm digging source gode
there's one trivial-looking conflict merging the defer-format branch to today's master
cradek: I notice that the standard nc_files "example" symlink confuses touchy
it should probably toss out everything that's not a file ending in .ngc
I'm not the least bit interested in adding a directory browser
I think the latest patch to threading looks pretty good. I tested it with accels of 5,20,90 and encoder down to 30 lines
was the problem at all related to having the first synchronized move be a short one?
I don't really profess to understand the problem or the fix..
yes that is what triggered the problem, but the real problem was once it wasn't perfectly synced, it oscillated instead of just catching up
now it syncs back up pretty fast, even though it still gets behind on those short moves because it can't go fast enough (fundamental limitation of the planner)
EMC: 03cradek 07master * rbe0dad0cd18f 10/src/emc/kinematics/tp.c: fix threading oscillation
hm, this says dircache is deprecated but it doesn't say what it's been replaced with: http://docs.python.org/library/dircache.html
probably with "just use listdir, it's not that slow"
Think back.. Remember when someone was trying to cut a cam using spindle synced motion... Would this help that?
on a lathe? no idea. the real problem with doing that is you can't cut at all those angles without live tooling
skunkworks is faster than me at pointers to the docs today!
heh - although it sounds like english is my second language again.
I really should read what I write.
I don't see any very badness words
This is what I am thinking of http://thread.gmane.org/gmane.linux.distributions.emc.user/15682
WHERE IS MY MOUSE POINTER?
EMC: 03cradek 07master * rb63d3d854b7f 10/src/emc/usr_intf/touchy/filechooser.py: ignore anything in nc_files that's not *.ngc
does it get rid of directories then?
(assuming so as it doesn't end in ngc ...)
not ones named something.ngc
... I guess
mshaver: I'm happy with my test results and I put that change on git master.
cradek: Much Better! See: http://www.mattshaver.com/problem/
files g76l0.png through g76l3.png
that looks nice. does it sound right?
sounds fine, I think you've got it solved
looks like your axis can't go much faster - are you sure it's keeping up?
the slope of the rapid looks about the same as the slope of the threading move
oh hey - referring to cutting non-circles on a lathe, I think that fastcut video shows multiple tools on the live spindle (when cutting hexagons)
don't you have that working yet?
not quite ;)
you get perfect flats when the tool radius is the same as the final cut radius
those speed are actually slow. it's a 3" long thread, done in about 3 seconds, so the ~ speed is 60ipm. It's actually that the retract move is slow.
you get two faces per tool, so you need N/2 tools on the spindle
what I don't understand yet is how they did the pyramid
mshaver: maybe I don't understand the plots then...
I see index reset, then Z moves negative for 3", then it moves positive for 3" at about the same speed or slower
yes, and the timing is 1 sec/box
the negative move is the threading move, the positive move is the retract
the Z axis on the machine can go 225"/min
so the retract move is not being made at the maximum rapid velocity
maybe you had your maxvel slider down
or feed override
I didn't even notice this until you brought it up...
maybe, I don't remember - and the machine is being torn down & shipped (I think)
no, but we don't have any extras - we ship as soon as we build them
heh, sounds like you need to build more
true. I won't be able to get another one for about a month, maybe longer depending on my schedule
well I'm too tired to fix any more bugs for a while anyway
great! we agree it's fixed! victory is declared!
too bad about 2.3 - not sure if this should be backported.
2.4 will be out in just 6 months, don't sweat it.
I think it's much better, but I'd hate to screw anyone up so late in the 2.3 cycle.
did you send your announcement?
nah, still sitting on it
I should, though
in six months I can test it some more!
in the code that runs the g76 cycle: Is the retract move done at the rapid rate?
well then, it's a mystery until next time
you could experiment with sim/lathe
cradek: is G10 L20 ... R ... accepted?
OK, I see it's not
oops. Just assumed it was (the quickref gcode sheet shows it...)
I clarified the "coordinate systems rotation" on the wiki page
emc/task/emctask.cc 321: interp_error: R not allowed in G10 L20
jepler: no, I do not see how it makes any sense to accept it. what would it mean?
I can envision it meaning "and keep my current point the same" but who the heck wants to figure out how to write that?
I should've been a sociology major or something.
but that's what the emc-users list is, right?
(which can actually be split into two interesting compound words)
- self.files = [i for i in self.files if i.endswith('.ngc')]
+ self.files = [i for i in self.files if i.endswith('.ngc')
+ and os.path.isfile(os.path.join(self.dir, i))]
cool, I couldn't figure out how to spell that
or os.path.isdir(...) ?
you want the files, ignoring the dirs
so no loading files outside of the nc_files dir? (or is that somewhere else)
< cradek> I'm not the least bit interested in adding a directory browser
I saw that
as I see it, the files you place in the nc_files directory are "in the control"
I use a network for this task
it seems that a very very simple browser would be OK though - if (isdir) then chdir and reload file list, else load file
nothing more complex than that
with a touch screen you could also easily drag files in and out
(assuming you can accurately touch the edge of your screen :) )
heh, they suck, don't they
yes, especially when the calibration software only runs under Windows
SWPadnos: this sucks
EMC: 03cradek 07master * r3de3360a910e 10/src/emc/usr_intf/touchy/filechooser.py: in listing, show files only