my normal machines spoil me for building software .. by comparison this single-core, 1800MHz "Sempron" is a dog.
yay, after months of neglect the little zenbot still works just fine
at least well enough to play with the modified pid..
sounds like really good results - cool, glad you tried that.
cradek: will you try it on one of your real machines?
jepler: they both have very high resolution so I'm not sure they're the best test cases, but yes, I can try it
I wonder what happens when you have velocity mode amp and use the velocity pid (using the velocity from mesa endoder)
if you use any D, it'll improve its smoothness
I was just reading the mailing list about the fest. I'm hoping to attend. I was going to ask if anyone is interested in further pursuit of that elusive Ubuntu-changes-related parallel port error (what, "...mode not supported, continuing anyway... etc.) I could probably get John to send his extra Pico-Systems USC board along if that would help, and I'll have one PC with a parallel port.
good - hope you can make it. I'm not sure what the parport issues are, but bringing extra parts is never a bad idea
cradek: that was weird, did you just try to send? (just before "good - etc.)
good - ... was all i've typed for the last 20 minutes
OK, must be an X-chat thing. Your send name appeared in front of my 2nd line of wraparound text. Odd.
I have two D50-based motherboards, and I think at least one of them has a parallel port
ah, they both do, except one of them has a header with 2mm spacing (silly)
Anyway, John still gets that parallel port error once on startup, and I think jepler said it had something to do with changes that Ubuntu made in how they handle the PP. I don't know if this happens only on Pico USC or if others see this too.
It would be nice to see that gone, but maybe there are bigger fish to fry.
do you have to run it twice, or does it whine but continue?
No, it just complains and writes a message box. It says "...continuing anyway." and it seems to because one start and it does run.
interesting - didn't know that would pop up a box.
sounds mildly irritating
I think it begins "PARPORT:..." if that helps?
"EPP mode not supported ... continuing anyway" or similar
not sure if it's EPP mode or "mode 4"(?) but yes, something like that.
It's not killing us, so let the triage begin. If we have to keep living with it, so be it.
I bet [the author] meant for it to be in dmesg, not a popup
I could start the BP2 and get you the exact text, if it would help you to think about it?
OK, here's the (transcribed) text in the Axis message box (red circle, white X):
PARPORT: linux parport parport0 does not support mode 4.
PARPORT: continuing anyway.
Hope that helps.
looks like it's printed with level ERR - wonder if that was on purpose
if there's nothing you can do about it ...
let's wait for jepler to comment tomorrow - I think it's his doing
OK, that's fine. I was only wondering how helpful it would be to have a Pico USC in Wichita, assuming I make it there.
hope you can make it - bring whatever's spare, just in case (especially if jon e can't make it).
I'm off to bed, goodnight
EMC: 03micges 07joints_axes3 * ref5dbaff8aa6 10/src/emc/motion/motion.c: Remove repeated line added during merge
the relevant commit is this one: http://git.linuxcnc.org/gitweb?p=emc2.git;a=commitdiff;h=a661c17
if there are real people for whom emc works after that message occurs, then we should simply lower it to a warning
(and it sounds like there are...)
EMC: 03jepler 07v2.4_branch * rdebc639a0dab 10/src/hal/drivers/parport_common.h: pico/7i43: don't show a non-fatal message as error
after this load error: http://pastebin.com/UDE15Q8j
I've got this error: http://pastebin.com/x4XjrRjA
and I can't load emc in any way
micges_work: you'll probably have to reboot to clear that error condition
I'll speculate that the problem was introduced in commit 44e7da328d5b7c038462e956694f6bd0f06fed10
.. assuming you're using master there
either way, there's something missing in the error handling that leaves the system in an inconsistent state where hm2_pci can't be loaded again..
jepler: I can't test pids today
but tomorrow or friday I will
micges: let me know if this patch fixes the problem .. http://emergent.unpy.net/files/sandbox/0001-hm2_pci-fix-unload-after-failed-registration.patch
does this documentation make sense and express what I mean? http://pastebin.com/Q9zSfdtE
yes I think so
I think that explains it well
it's too bad the pin can't be made an output when it's not in use, for monitoring purposes
actually it basically is
when a pin is created it points at its dummysig
each time its function runs, pid puts the estimate on the dummysig and then uses the value from the pin to compute the D term
unlinked -> old estimate is used and appears on that pin as though it was an output
linked -> external estimate is used and old estimate can't be seen in hal
it's a bit of a nasty trick..
right. but you can't connect it to a pyvcp meter, for example
yes, you have to know to meter/scope the pin directly, not connect a signal and scope it
anybody want to stop me before I push this pid change?
nope, it needs the light of day
give it a shove
EMC: 03jepler 07master * r8702094f3779 10/src/hal/components/pid.c: pid: read command, feedback only once
EMC: 03jepler 07master * r44586b831e1e 10/ (docs/man/man9/pid.9 src/hal/components/pid.c): pid: accept external command-deriv, feedback-deriv
hmmm. the "read inputs only once" patch isn't complere
only the error calculation was replaced, not derivative etc.
are you using stock gnome wm on production controllers?
my mill has regular gnome with touchy set up to auto start
I mostly use the gnome toolbars for shutdown, and very occasionally for file browsing (using places/home ...) to do something like view the tool table
i'm installing ubuntu for end-user so asking for 'best practices' :)
so let gnome be there...
ah - IMO best practice with ubuntu is to not change anything that will make finding help online harder for end user
yes :) and that's why i don't install dwm with preconfigured emc start :)
SWPadnos: hmph, now you tell me
heh. well I didn't see it before :)
finally i have working python bidings for camunits framework! now i may add camera widgets with glade :)
EMC: 03swpadnos 07master * rdf3f3609bd99 10/src/hal/components/pid.c: Remove a couple of accesses to the command input pin, using
I suppose I could have tested that first
somebody will. ;)
heh. destructive testing ;)
it compiles ;)
is there a.y packages for HEAD?
or i пeed to build theм мyself?
I did compile test it
psha: your text is screwy here
I think the buildbot makes testing packages, but I don't know if they are in a repo
psha: buildbot regularly builds packages from git master, if that's what you mean
you'd have to download and install manually
[18:14:59] <cradek> http://emc2-buildbot.colorado.edu/~buildmaster/
I am seeing "or i \u043feed to build the\u043c \u043cyself?" and "tha\u043fks"
<psha> or i пeed to build theм мyself?
that looks normal to me
me too, and that's what I see (in Chatzilla)
13:20 < skunkwork> <psha> or i \u043feed to build the\u043c \u043cyself?
cradek is probably seeing the real thing in irssi
looks like caps are screwey
cradek: sorry but .y keyboard is partially killed by kid
This... 2010-10-27 11:12:22 psha: or i –øeed to build the–º –ºyself?
letter betwee. B a.d < are .ot worki.g
[18:26:30] <Jymmm> http://codepad.org/pjl2awcz
so i'. replaci.g the. with cyrillic :)
on the commodore 64 the space bar often quit working (until you took it apart and cleaned out the ground-up plastic dust) so I'd sometimes give up andtypewithoutit
and here is my only working keyboard
connected to testbed
here most of the odd letters are showing as dots, but there's an "m" somewhere back there that is actually a cyrillic character.
at least I know what's going on now
SWPadnos: thanks for fixing that fix
I hope you got it right, otherwise everybody's going to blame my pid fixes
sure. is it fixed?
i was replacing m and n either with dots or with cyrillic :)
psha_ is now known as psha
First, rewinding head to replay your work on top of it...
Applying: pid: really read command only once
Using index info to reconstruct a base tree...
Falling back to patching base and 3-way merge...
No changes -- Patch already applied.
SWPadnos: neat! your change was exactly the same as the change I'd made and committed locally .. and git figured that out and did the right thing
EMC: 03jepler 07master * r6f2907912b9e 10/ (4 files in 4 dirs): axis: optionally use a blend mode for preview plot
jepler: have you tried 1px backplot lines?
psha: I looked at your screenshot and tried a few different things but I haven't found anything I'm thrilled with..
so let's trust your artistic taste
checking for sarcasm ... indeterminate
i honestly honor others taste since i'm horrible in desiging interfaces/look-n-feels...
jepler is good at look/feel - I trust him too.
thanks for the votes of confidence
don't let it stop you from showing me your ideas though
ries_ is now known as ries
jepler: Thanks for making that change to v2.4_branch on the parallel port error. We'll look forward to trying it when 2.4.6 comes around. I noticed the CIA-2 report mentioned "...pico/7i43...", those are the only trouble spots? No one else uses EPP? I'm surprised.
well pluto does, when it feels like working
KimK: it doesn't affect hal_parport users (no epp used there). It affects pluto users but I didn't mention them because I wish there were none.
but those are the significant players anyway
Ah, OK, thanks. I don't know anything about Pluto. (And maybe I shouldn't, from the sound of it, lol?)
it's a board designed with about 1% of the skill and care of the epp boards from pico or mesa
for 1% of the skill and care, it runs on about 30% of machines - that's not too bad a payoff
OK, thanks, I'll Google it sometime.
googling "pluto" should be real effective.
I was trying to find good examples of "spotting" earlier (a type of metal decoration)
google was very little help
Ha, yeah. (But wait, now that it's 'not a planet' anymore... oh, nevermind)
Re Pluto not a planet anymore, funny image: http://www.mathiaspedersen.com/portfolio/3d-portfolio/poor-pluto/
regardless of what it does, surely naming a program "the same letters as rm, but backwards" is a bad idea waiting to happen. http://kitenet.net/~joey/code/mr/
(bonus points if 'mr' uses the -r flag; I haven't looked)
maybe -m ?
to keep it consistent :)