jepler: You are not incompetent, it is fixed :)
but you can see that I disobeyed my rule to test the change before committing it :(
The good teacher shows what not to do
after switching to the lower speed grade and playing with P&R options, I can get up to 166MHz on my design
that's probably fast enough :)
SWPadnos: does it seem backwards to you that -4 is the slow grade and -5 is the fast one?
hmm. yes and no. it depends on how I'm thinking of it at the time
"higher number = better" vs "number = propagation time"
I tried to read peterw's fpga cpu design (dumbass8?) but that was all waaaaaay over my head
it was funny that his IRC client was still set to "PeterDumbassW"
I didn't notice
he changed it just after connecting
I haven't looked at the processor yet
It's only 300 lines so how hard can it be to understand?
that's a good theory
[00:30:21] <jepler> http://pastebin.ca/768962
as though you need to be distracted from whatever it is you're doing
'THE BAD NEWS: pipelined processor with no locks'
heh - I'm concentrating hard on what to do with all the resources I have in OGame at the moment ;)
ie, goofing off
hmmm. that first data hazard is a doozy
"Stored data requires 3 instructions before fetch"
peter isn't going to let me off the hook for that comment is he
particularly as I don't see a NOP instruction
the dumbass8 processor - from PeteW and JohnK :)
jepler: JMP <here+1> == NOP
not in a branch delay slot I'm guessing
except that it always executes two instructions after a jump
there's probably something that would work
there are no delay slots
SWPadnos: well, whatever you call those two instructions after the jmp
oh, if they weren't executed I'd call them delay slots. I don't know what they are if they do get executed :)
hm, I don't see any way to transfer the saved pc counter so you're limited to one level of subroutine?
says two in the comments
I see that
vhdl sure is verbose
and then it has constructs like this:
when '0' => idn0 <= idx;
there really should be a 'downto' in there
I mean on that particular line
it's too terse
oh, that's not what it means
also, why is everybody ganging up on idn0?
here's a good one:
if accum = x"00" then
zero <= '1';
zero <= '0';
-- operate instructions
constant nop : std_logic_vector (3 downto 0) := x"0";
it's obvious what it does, but it sure lookd funny assigning 1 to Zero
oh good, there is an explicit nop
that's why it should be called Zf or something
I wonder hw much logic it adds to have an internal flag to skip the 2 instructions after a jump
called "Branching" or something terse
jepler: sorry, juste a comment,
Some error message displayed not translated
ie: emc/task/emctask.cc 312: interp_error: Selected tool slot number too large in place of Numero d'outil trop grand
yes, that's a known but that nobody's gotten around to fixing. http://sourceforge.net/tracker/index.php?func=detail&aid=1650938&group_id=6744&atid=106744
ok sorry, time to bed here, good night all :)
03tissf 07v2_2_branch * 10emc2/src/po/ (fr.po fr_axis.po fr_rs274_err.po): French translation update
hey Ray. how goes the stepper thingie?
03tissf 07v2_2_branch * 10emc2/docs/src/ (5 files): French translation
03tissf 07v2_2_branch * 10emc2/docs/src/hal/ (intro_fr.lyx tutorial_fr.lyx): French translation
is there a scratch.txt somewhere for testing cvs access?
cvs [commit aborted]: could not open lock file `/cvs/emc2/,TODO,': Permission denied
does this mean I have an anon checkout?
there is another module (not emc2) where there is a scratchpad
I think it was in documents/
hmm, maybe I'll do some playing around today only and fix cvs later...
cvs is not supposed to as about passwords right? (when the .ssh keys are configured correctly)
when someone (cradek?) is around who can update my ssh-keys, let me know! (I've switched machines again...)
03tissf 07TRUNK * 10emc2/src/po/ (fr_axis.po fr_rs274_err.po): French transaltion update
you can copy ~/.ssh/id_* to the new machine
yeah, unless he reinstalled, and lost the old id's
03tissf 07v2_2_branch * 10emc2/src/po/fr_axis.po: French translation update
can HAL pins be RW? or is RW only for parameters?
I think for pins it's IO
(but yes, they can be bidirectional)
[13:24:17] <alex_joni> http://www.linuxcnc.org/docview/2.2/html/man/man3/hal_pin_bit_new.3hal.html#ARGUMENTS
is there a conceptual difference between RW and IO ?
and we plan to replace params with pins
there is no real advantage for a param :)
yes, I'm trying to work on PID-tuning via pyvcp right now
I think a tree-like explorer gui for pins would be nice for halmeter/scope in the future (there are a _lot_ of pins on a big machine)
check out halshow
right, something like that.
halshow with wires!
we could take a cue from msn and call it 'halshow Live Edition'
is halshow built in the standard package? bin/halshow?
it's in tcl somewhere
03jepler 07v2_2_branch * 10emc2/configs/sim/servo_sim.hal: the linkxx -> net conversion had broken the servo_sim configuration
03jepler 07TRUNK * 10emc2/configs/sim/servo_sim.hal: merge from v2_2_branch: make this config run again
jepler: this is the 0 0 0 to 1 2 3, and the wcomp.9 to wcomp.8 ?
I could have sworn I tested that after I changed it, but clearly I didn't...
jepler: do you remember anything about RW pins in pyvcp?
no, should I?
hm, I just want to use a scale widget, but have it initialized to whatever the signal value already is
now it forces the signal value to 0 when it starts
searching, I found this -- no idea who wrote it. http://pastebin.ca/337671
but that doesn't fit your needs
it's an unavoidable feature of the Tk slider that it sets the variable to some undesired value when it is first displayed
axis has some workarounds (the feed override and spindle override sliders are affected by this) but it looks like they just force it to 100% at start ..
would a numerical control be easier in this respect?
I wouldn't be sure without testing
I'm trying to work on PID tuning through pyvcp. would be nice if the PID values wake up with what is in the ini file and then allow the user to tweak them
darn, awallin left. http://pastebin.ca/769581
I have dual personalities...
thanks, I'll try that.
03awallin 07TRUNK * 10emc2/TODO: m5i20 todo list
yay, my vmware ubuntu still has ssh configured right!
awallin: then it's simple to replicate that
alex_joni: just move everything in ~/.ssh ?
did someone work on include-files for pyvcp already?
how are things?
Good. Been working on a pick-and-place using HAL.
How you and the family been?
great. the kids are getting taller and taller....
Ben is a lot taller than me now
I remember that as a real difficult thing for me.
You working with EMC these days?
not at the moment. I've been working on cam stuff more than anything
learning more about python
still working with apt
but, I'm still very interested in emc
03awallin 07TRUNK * 10emc2/src/hal/components/pid.c: change P,I,D,FF0,FF1,FF2 to pins to allow tuning with a pyvcp panel
I see cradek has done some work with probing lately
Have not had a chance to see it but sounds good.
jmkasunich__ is now known as jmkasunich
awallin: previous configs shouldn't be affected by the pid change, right?
seems to me that as long as the param values were being set by "setp param [AXIS_0]P_GAIN" it will still work
setp switches from "set parameter" to "set pin" if the param doesn't exist and nothing is connected to the pin
jmkasunich: I know the theory .. was a bit too lazy to test it out :)
regarding the 'axis.N' component, will there be an axis.N >and< a 'motion' component for each joint?
axis.N is just the name of some pins sticking out of motmod
i just read thru 'man 9 motion' to see what was created, and the pins are named axis.N >and< others are named motion.blah, not motion.N.blah. are there 2 comps created? is there numbering on motion?
motion.blah is only one-fold
there is only one component called 'motion'
it exports the motion.blah stuff
and besides that it exports joint specific information, called "axis.N"
so 2 comps, motion may already exist, right axis is joint specific, thanks
not really 2 comps
it's just one comp, creating different pinnames
i'm trying to make linkable symbols for gEDA, and thats visually troublesome... visually i'll make 'motion' separate
are any of the documents or drawings at http://home.att.net/~jmkasunich/EMC_Docs/EMC_Home.htm
I'm going to be changing ISPs, and the hosting will be going away
should I bother copying the stuff elsewhere?
jmkasunich: I think so
was afraid you'd say that
if you want, I can duplicate it to linuxcnc.org
that would be fine by me
I haven't maintained that page in years
[20:17:43] <alex_joni> http://www.linuxcnc.org/content/view/42/13/lang,en/
moving pictures now
jmkasunich, if ou want to copy it to your DH account, that's fine by me
having it under linuxcnc is better IMO
oh sure - just backing it up somewhere first is a good thing
then moving stuff around where we want it
it looks like alex already found a place
hmmm. if I do it, may I rip off your images for a presentation on EMC?
be advised that they are somewhat out of date
proper attribution of course
SWPadnos: only if you update them :D
damn. I may have to update them
darn.. too late
[20:27:11] <alex_joni> http://www.linuxcnc.org/content/view/42/13/lang,en/
they were drawn with easycad, but that was several computers ago
jmkasunich: can you please look if I didn't miss anything?
but to do that, I'd have to figure out how it works now ;)
if I can find them, I'd be happy to convert to dxf to allow for future updates
SWPadnos: the first one is quite good, (only the IO controller is a single thing now)
not divided in 3 anymore
no biggie. I may just redo it in OODraw or something that runs natively
and the spindle control is in motion now
alex_joni: looks identical to the original
only a little formatting needs to be done, fro mwhat I see here
the image widths need to be reduced
ditto for the second one
yep, I think they're pretty darned close still, though some of the names may have changed
that's why I said we should definately keep them
the widths are what they are (gifs created from the vector originals)
I can make them smaller..
but it won't read as nice
but the image <width=xxx> needs to be reduced
I guess since you can click and get the full size ones it doesn't matter
on my machine, the images are already compressed, but are wide enough that the text above them ends up overlaying the right side panel (the poll)
yeah. if I make my browser window narrow things get messy fast
hmmm. this isn't exactly "narrow" - it's on a 1280x1024 monitor, but not quite full width
I think the center pane width is determined by the widest fixed width thing in it, which is the drawings
1600x1200 here ;-)
I reduced them
I could load it up on the 3840x2400 monitor and see what happens ;)
the internet would be 5" wide instead of 3 then :)
yeah, the full version would be great for that
the reduced ones are ugly, but I'm not sure what else can be done
jmkasunich: you still get the big ones when clicking
about the only thing would be to use something like imagemagick to make a better small version
so I'll leave it at that for now
jmkasunich: one day you'll have to learn how to use joomla :D
* jmkasunich runs away
some things work surprisingly well :D
(like copy paste from another browser into the new content field)
it even preserved formatting and links
03jmkasunich 07TRUNK * 10infrastructure/farm-scripts/index.shtml: adding the 2.2 branch
23:49 < CIA-18> jmkasunich TRUNK * infrastructure/farm-scripts/index.shtml:
adding the 2.2 branch
oh - I didn't see that, thought CIA was sleeping again
I was disconnected
you were having your little hickups again
I sure hope that changes when I get my new ISP
btw.. they are like clockwork :D
03compile-farm 07Ubuntu 5.10 (breezy) non-realtime (2.6.12-10-386) * 10emc2.2branch/: build PASSED
I should be getting connected on Wednesday
I thought it was a little funny that we got the commit, and 1 minute later the disconnect message
[22:02:55] <alex_joni> http://www.linuxcnc.org/compile_farm/
the disconnect isn't visible (to you or to me) until after the IRC timeout
it's starting to populate :D
the most time consuming part is getting the new checkout
can't you copy the whole dir over? once it's checked out?
from one VM to another?
I suppose thats possible, but more trouble than its worth
I suspect that works faster then cvs up
it only takes a few minutes
and by doing a fresh checkout, I know all the CVS/foo is correct
I wonder if I should eliminate emc-2.0 from the farm?
does it use up lots of resources?
some vm disk space
disk space, which I believe I have plenty of
and cpu once in a while
(if? there is a commit there..)
a few seconds to do the cvs up each time
no compile unless there is a commit
(last compile was in Aug for one slot, and in Feb for the other two)
wonder why the big difference
I think the last commit was in Feb
I may have restarted the farm in Aug, and forced a compile
nope there are lots of compiles on the bdi platform
[22:11:05] <alex_joni> http://www.linuxcnc.org/compile_farm//emc2.0branch_slot6_history.txt
lemme get 2.2 working on all slots, then I'll investigate
not really something to lose sleep over
on that note.. I'm off to bed
good night all
03compile-farm 07BDI-4.51 (126.96.36.199-rtai) * 10emc2.2branch/: build PASSED
third one building now
the 4th one is my other machine, which (a) isn't running at the moment, and (b) isn't connected to the net the last week or so since my router started misbehaving
the new ISP's new DSL modem is supposed to contain a built in router, so I'm not gonna do anything in the meantime
moving 2000 lb of shingles onto the roof is not easy.
is it all sheathed?
I suppose I should have guessed that was the case.
they have forklifts for that
shingles go on tomorrow
roofer coming tomorrow
03compile-farm 07Ubuntu 5.10 (breezy) realtime (2.6.12-magma) * 10emc2.2branch/: build PASSED
so you were carrying bundles up a ladder?
or did you work out some labor saving thing?
the first two, yes
after that, I figured I had to pick lever, pulley, or wheel to help
doesn't matter, you still do all the work unless you use motors
so, pulley it was
SWPadnos: the motor was attached to a large Ford
after doing my mother's porch manually, we hired roofers to come and do our house
SWPadnos: carrying up ladder means moving 50 lb bundle of shingles + 190 lb bundle of human flesh up to the roof, and then taking the 150 back down again
I weigh about the same as two bundles of shingles, so I had devised a system where I would climb the ladder and ride the rope down, lifting two bundles
175 - who are you kidding? :)
well, I'm 190
but, I chickened out (i.e. got smart) and used the truck instead
the hard part of the system must have been attaching the pulley
I made a tripod/triangle that went above the peak
you need an assistant on the roof to move the bundles over
with the rope-riding scheme
yes it would have been harder, but possible, with just one person
or you tie the end, and climb the ladder
these bundles were 80-85lb
hmmm. that sounds bigger than the standard ones
usually you use 3 bundles per "square"
they seemed plenty big.
which I believe is 10x10 feet
its been a while since I lugged shingles - I pulled 50 lbs out of my rear
I think they said 245lb/"square"
hm. they could have been 85 poounds, but I don't think I'm that strong
I realised shortly after I hit enter that "out of the air" would have been better
we must have been using cheaper shingles
are these asphalt shingles or slate or something?
I'm not sure
if they're slate or something you'd know
they're the standard "kinda flexy with little pebbles in them" kind?
(twice the price and special skills to install)
ok, they're asphalt
but they don't burn
what is slate?
right - sorry
[22:41:25] <jmkasunich> http://en.wikipedia.org/wiki/Slate
has a photo of a slate roof
my parents house used to have slate
that's one ugly roof they have in that photo
but it's probably 200 years old
I think the roof on my parents house was about 75 years old when they decided to replace it
I haven't seen that around here
unless you have houses approaching 100 years old you won't
heh - here's why I've seen it, I think: http://www.vtslate.com/
there was a pile of shingles across the street when I was growing up
we have lots of houses that old in lincoln
someone had probably replaced their roof not many decades before
wow. those specials are pretty darned good
$100-$150/sq. is only 2-3 times the cost of asphalt
I think. how much are normal shingles these days?
yeah, but that doesn't include labor
no, but it also doesn't include replacing the roof in 10-20 years
I bet installed price for slate is 5x or more compared to asphault
cost per year is another story, but few people think that far ahead these days
if it's uncommon, yes, but here it's only ~3x, including the shingle cost
it's still a big chunk of change though
we lokoed into standing seam metal, slate, and asphalt (as well as DIY asphalt), and the range was $2000 or so (DIY) to $12000 or so (metal)
good night all
yay - finally finished my trash can
heh. what does that mean?
I needed to make a lid
the can is an antique 10 gallon pickle crock
the lid keeps the dog out
10 gallons. that's kind of small for a pickling crock, isn't it?
I think they made 10 thru 15 or so
I think we used to make sauerkraut and pickles in 30 gallon crocks
we used to have a 12, it went away in the divorce
the 10 is big enough
but they may have looked biugger because I was 6 years old
the 10 is 16" OD and 17" tall
plenty big for a kitchen can
ok, ours were considerably bigger
maybe 20" across or a bit more, and 30-36" tall
crocks are expensive these days
I was searching for a while, and not finding anything under $100
I got luck and got this one for $48
and with pickles only $2/gallon at Wal-Mart, who needs them ;)
wonder which is more expensive, slate or solar panels
solar, I bet
100 square feet of slate is a few hundred $
a 1 square meter solar panel (~10 sq.ft.) is more than that, I think
oh, thought you said $150/sq. foot
no, per "square"