EMC: 03micges 07iotask_remove * r25b1c0c655c0 10/ (327 files in 96 dirs): Merge branch 'master' into iotask_remove
ries_ is now known as ries
cradek: If you see this I wanted to let you know that I rebuilt usbtouchscreen.ko and got the touchscreen to respond after rmmodding usbhid and then hotplugging the USB cord. I think I can use usbhid quirks to get that driver to ignore the touchscreen and let it be controlled by usbtouchscreen instead. The _real_ problem stems from the fact that the touchscreen reports itself as an HID, but really isn't HID compatible.
spectacular - congrats
It still doesn't work right, but once I get done I'll do a whole big writeup on the complete process.
but it moves the pointer?
(what a pain in the ass badly-behaved USB devices are)
eh, it seems to register all touches in the lower right corner of the screen, repeatedly opening the trashcan
I guess that's something.
I think that's related to another problem where the driver thinks there are two touchscreens instead of on (shown by xinput --list)
It's a multi-problem type of situation...
anyway, time to pass out for the day! Thanks for all the help!
morfic- is now known as morfic
ries_ is now known as ries
JT-Work_ is now known as JT-Work
does this make sense? http://sherline.com/CNCmenu.htm
what is an incremental mode preset?
you mean that they knew about a bug in february but didn't tell us? nope, doesn't make sense.
SWPadnos: they mean increment settings, I bet
ok, jog increment settings
alex fixed it (??) in jan 2010
I wonder if someone should email them
"hey, this was fixed the month before you posted this message, so if you update (back-date?) your version of emc2, it'll be fixed" ...
"unless of course it's not actually fixed for you, in which case it might be nice for you to tell someone"
hard to say whether alex's fix is what they report (but it sure seems likely)
knowing the interesting change is a5133a5231f7, how can I tell which tagged release versions that's in?
cradek: well there's probably a smart way
but the dumb way I do it is: gitk -1 a5133a
click the commit
look at the Follows: and Precedes: lines
Precedes: v2.4.0-pre1, v2.5.0-pre0
so all of 2.4, none of 2.3
to see what was tag before commit you may use git describe
$ git describe a5133a5231f7
psha: that's not really what I want - I wondered which releases follow (or which branches that change is merged into)
hm, I guess it also may have been cherry picked in some release branches
i see :) so i've stated that it only shows tag _before_ :)
cherry pick is impossible to track as i understand git
since it's just reapplying of commit
Is Ray still in contact with Sherline? If not, is anyone? I ask because it's probably a good idea if we make sure that they are shipping the latest & greatest EMC2, and that their users know how to get upgrades, fixes and support. I know it's not really our job, but Sherline is a significant portal through which many people get their first exposure to EMC. It would be "a good thing" if their experience was as positive as is possible.
Does anyone from Sherline ever get on here and talk?
years ago there was a guy who was routinely in contact. he left sherline and I haven't seen anyone since.
Was that Mike <something>?
The guy who was out in Ann Arbor years ago? Dark hair, youngish?
yes mike something - I never met him in meatspace
He was somewhat, ...volatile.
no one else eh? Any word from Ray lately?
ray drops me an email once in a great while, but he hasn't been on irc for ages.
buildbot is up?
I think seb is working on it at the moment..
when new packages will be ready i'll report how new rendering in axis works in real environment
"With cutter compensation in effect, some dwell will occur for blocks that are executed in less than .25 seconds."
I thought block processesing was .25 seconds for all moves ;)
our cutter comp is now officially better than this control from 1983 had.
we get all the cases right that it does, plus two that I see that it gets wrong
and we probably only use 100x the ram and 1000x the cpu power to do it
exactly.. remember the good old days where the os fit on a floppy?
The part program storage area is equivalent to 100 feet of NC tape and is non-volatile.
by my experience that should be "and is sometimes non-volatile"
"and is as non-volatile as NC tape stored in a warm, damp place"
"... stored in direct sunlight"
actually - I think the test tapes we got with the K&T are metal
cool, it could do a grid ("frame"?) of holes with one block
G191 X Y Z X Y Z P P F
maximum control limit, coordinate words: +/- 838.8607
aha, that's sign plus 23 bits
3 byte words!
cradek: my connection is a little shaky today - do I understand that you've implemented all the Bridgeport mega-G-codes in the interp? :)
sorry to inform you, but your grasp on reality is apparently a little shaky too
(someone asked me about adding filleting - I remembered boss8 did it, so I looked it up)
but you did say that our cutter comp handling is equivalent?
not equivalent to filleting
I was just commenting that our cutter comp is better than boss8's :-)
but no gouging errors for too small inside radii?
I don't understand the question
boss8 would not balk at a sharp inside corner with comp on
emc has handled that for some time now
*since cradek fixed it...
really. I did not know that.
where have you been?
not in front of a machine obviously!
April 2009, EMC 2.3.0 release note: "new cutter compensation algorithm handles inside sharp corners and does not require complicated entry moves"
so saying "for years" would have been a bit hyperbolous...
actually, I've been trying to get this touchscreen to work - which once i made some adjustments to xorg.conf.d/10-evtouch.conf and blacklisted usbhid is now almost sort of working
it's just a little off tracking a stylus - need to get the numbers right and I think it will work
anyway, back to work on that...
cradek: polar coordinates can be applied to G2/G3 ?
micges: yes you can specify the endpoint in polar
JT-Work_ is now known as JT-Work