I'm glad that mystery is solved
[Global Notice] Hi all! This morning (6:00-7:00AM UTC), we will be restarting a major portion of the freenode IRC network. More information regarding these updates is online here: http://freenode.net/news.shtml
Further questions can be directed to firstname.lastname@example.org or an on-duty staffer in /stats p
yes indeed - good catch, jmkasunich
the discussion between you and jeff is what made me suspect some precision loss somewhere
we suspected it as well, but didn't notice the cause
are you still subscribed to the Geckodrive yahoo group?
"subscribed" yes, but I never had mail delivery, I read it on line
and I haven't been lately
ok. somebody posted a question like "how can I get ethernet working right?, and I responded that I was considering building an ARM board to replace the rabbit
interesting discussion ensued :)
the original subject is "GeckoMotion Over Ethernet", from 12/1
* jmkasunich reads
the main reason I was thinking about my own processor board is because there's no Linux devkit for the Rabbit (that and it's $300 or more)
seems Mariss got a little defensive
a little :)
I didn't read anybody as really bashing his choice
he was being attacked from several sides though
yeah. I guess he was rather blunt
a bit ;)
I thought the 5ms block processing time was excessive
certainly for HSM, probably not for hobby machines
lack of 32 bit math operations is the real issue
it's a big one
just takes more cycles
I used to like Z-80s, but then I started using 16 bit machines
I liked them, until I started using 32 bit machines
there is no going back
unless the task is truly rudimentary
I never really liked the Z80 much - it was pretty fast for the time, but cycle counting was such a PITA
it's pretty obvious how they got to the X86 from the Z80, when you look at all the special-purpose uses of the "general purpose" registers
and the lack of general-purpose registers :)
you mean to the X86 form the 8080/8085
Z80 -> 8080/8085 -> 8088/8086 -> x86
not if I recall it correctly
z80 is post-8080
8080/8085 were first (intel)
hmmm - that's the history I've seen
z-80 was zilog's improvement on it
oh - was the Z80 Zilog's "new and improved" 8080?
well, you can see the common roots then ;)
crap - I'm going to need to reboot soon. windows explorer is fubared
(insert smartass comment here)
I'm sure Vista is better though
(how was that?)
not bad for a first try
I suppose it would give me a chance to clean part of my desk or something
or set up my server
if anyone wondered, cat paws do run trackpads
oh good - they sometimes run mice as well
our remaining cat is too lazy to get up on the bench
or table, or bed, or anything else more than 12" off the floor
hmmm - has he lost weight?
well how often do you jump on things twice your height?
I'm not a cat
ok - our cat isn't expected to live beyond a week or two - one of the symptoms he has is general weakness
(we decided not to put him to sleep)
it's a symptom of renal failure, which is common in older cats
yeah, we've been there
yep - most cat people probably have
our cat's pisser works just fine
he has an annoying habit if getting all four feet in the litterbox, then hanging his ass over the edge and peeing on the floor
at least he's (sort of) trying ;)
"but I was in the box" ...
he's not the sharpest pencil in the drawer, but he's sweet
our guy has never been too cat-like. he's been more or less a vegetarian for the last several years
he also manages to get stuck to the carpet from time to time
(claws, not astatic)
thats bizarre (the veg part)
what does he eat?
yeh - he likes the smell of meat / cheese, but never actually eats it
obligate carnivore and all that
he eats catfood, whatever that is
grain and meat byproducts probably
he'll eat tuna / beef / chicken catfood, but not cooked or raw "human food"
well, that just shows he's a smart cat
but he does eat things like peas and string beans ...
human food isn't good for felines
ham, beef, chicken, fish, cheese, milk - those aren't bad for a cat
well, milk can cause the runs
its not a complete diet tho, unless you include some organ meats, etc
"not good for" is probably an exaggeration
sure - not complete enough
but all other cats I've known would yell "Look, Elvis!" and then snatch food off your plate
unless they just take the food instead
and they could all detect a gram of protein from roughly a mile
hey pete - long time no see
I just made some mods to the motenc dirver
but what is BWidget?
its a tcl/TK thing
what package is it in?
ii bwidget 1.7.0-1 A set of extension widgets for Tcl/Tk
it's a tcl/tk UI widget set
ok, let me apt-get
"bwidget?" means "I think the package is bwidget" ;-)
isn't that one of the dependencies of emc2-dev?
we use the tree widget from it for the config picker that opens when you start EMC2
I'm still on an old debian box
I'll try emc2-dev
and its also used by halshow and maybe a few other things
you may not be able to isntall that on a non-ubuntu system, though I'm not sure what all the dependencies are
does anyone have a motenc setup in testable condition?
I think none of us has a motenc setup at all
except Ray probably
what location is emc2-dev package at?
theres the one in Galesburg ;-)
I think bwidget gives the tree widget
but thats a bit inconvenient
I guess that's testable ;)
I need to update my apt-sources
what do I need?
but it's not dapper so ...
or breezy ...
cradek: will that work on a plain debian box?
I don't know
doubt it will break anything, but all the deps might not match what's available
can you check the emc2-dev dependencies - I don't think emc2 is one of them, and the other packages are all dev stuff
(like gcc and the like)
my apt-get couldn't even find emc2-dev with the sources I had
are you talking to me?
err - yes, I think I was ;)
no - the emc2 repository is where emc2-dev is found
cradek, yes, I was talking to you
hmm, says needs gcc-4.0
(just so I'm not confusing everyone)
is that really a requirement now?
SWPadnos: I don't understand what you mean then
petev: no, you can use any compiler
petev: do you have a RT kernel, or are you using the non-rt simulator?
just wondering if emc2-dev includes only development tools, or if it would also pull in emc2 and the RT kernel, etc.
petev: you should use whatever you built realtime/kernel with
if you are building for RT, you need to use the kernel compiler
(what he said)
emc2-dev won't install without gcc-4.0
SWPadnos: are you talking about install/run dep, or build-dep?
then don't use emc2-dev
petev: you're going to have to wing it
ok, let me just try and install bwidget
cradek, no, there's a package called emc2-dev
it depends on a lot of build tools
didn't someboduy like you make the package?
I know what it is - I don't know what you want me to do to it
oh - I just can't look at the list of dependencies right now - that's all I was wondering about
Depends: g++, gcc-4.0, linux-headers-2.6.15-magma
cradek: do g++ and gcc versions track?
Build-Depends: debhelper (>> 4.0.0),linux-headers-2.6.15-magma,g++,gcc-4.0,make,libc6-dev,tcl8.4-dev,tk8.4-dev,libgtk1.2-dev,pciutils-dev
(IOW, should it depend on g++4.0, instead of just g++_
man cradek, your going to make me have to install an ubuntu system
if your PC can handle it, it's not a bad thing
however, if you have a nicely tuned system, then any changes are a bad thing ;)
but we should _not_ be forcing people to discard perfectly functional debian systesm
yeah, but it's all nice and pretty now with my touch screen drivers and all
petev: do you have a RT kernel in that system?
what version of X?
I installed bwidget and so far so good
building a completely new tree
so the part about "linux-headers-2.6.15-magma" is wrong anyway - you want the headers that go with your kernel
emc2-dev is a non-issue. I thought it might help, but it won't
you don't really want that repository - it's named "dapper" for a reason
is it a big deal for the encoder interface to have more than what canocical calls for?
what do you have in mind?
I think the canon is a minimum, not a maximum
just some stuff that was already there
like the count reset pin
I would probably agree, but I hope for some consistency
reset is a "reset now" thing? (as opposed to index, which is a "reset when you see the index")
I have no problem with a reset pin
ok, just commited
in fact, someday (version 2.2 maybe) I'd like to revise the canoncial;
will need testing
mac will like that
instead of "index-enable", have two: "reset-enable" and "capture-enable", and have "capture-counts" and "capture-position" pins as well
yes, that would be nice
both of those enables would work like the existing one, as far as the handshake goes
(set true externally, when the index arrives the desired action happens and the pin is set false)
I hate to ask this...
if I command S500 and then S-500, is a properly configured machine going to do a sane thing (reverse without destroying anything)?
keyword "properly configured"
ok good :-)
I thought that was the answer
because the configurator will have designed it to do the right thing
which might include a limit2 block to control the slew rate of a vfd, etc
yes, like I did
rigid tap is ...close...
how usefull is a 5V 4A power supply in a nice small package?
seems like that would often be nice on the bench
3x5x1" similar to a laptop supply (IEC connector on one end for AC, cord on the other end)
if you put it in a nice box with an ammeter and banana jacks, I bet you'd use it a lot
hard to say
I have a nice variable supply already
and another even nicer one that just needs a little work
I found a space to store the 5V ones tho
(there were 3 of em)
hmm, my approach is wrong
anybody need some KVM cables (nice ones!) http://jmkasunich.dyndns.org/pics/kvmcables.jpg
I think I'm going to have to allow real path reversal
what were you trying to do?
let them program down and back up as two moves?
just have two synced moves, one down, one up
well at the TP level - coming from gcode can be anything
the first move will never end at the right spot unless you have a position loop on the spindle
I'm letting it go past, that's solved already
(it will reverse at the programmed depth, so it will travel a bit past)
but the problem is I have to let the spindle encoder count up and back down and follow it -- if I unsync I have to wait for another index, which I can't with the tap in
so while going down, Z = K*spindlepos, when you hit the desired depth you set spindle speed to negative
when spindle reaches zero, you start the next move, with Z=-K*spindlepos
yes but the spindle encoder is counting down now
so all the calcs are wrong
(threading with the spindle reversed has never worked)
is there some variable that goes from 0 to 1 (or something like that) over the length of a move?
0 to length, yes
what is that var called?
during move #1, progress = K * spindlepos
when progress = length, stop spindle ASAP
after it stops, begin move 2
during move 2, progress = length - (K * spindlepos)
_if_ you could stop on a time, then progress would start a K and go to length as the spindle unwinds
but if you overshoot the first move, then progress will be negative at the start of the second (by the same amount as the overshoot)
I think that can be fudged
this requires that K and length be the same for both moves, not an unreasonable think to enforce
man, I fscked up that statement (3 lines back)
I think I know what you meant
_if_ you could stop on a dime, then progress would start a ZERO and go to length as the spindle unwinds
I think you're saying I have to fudge all the numbers and minus signs...
if you're gonna treat it as two moves in series, yes
the other option is to have one move and allow backtracking along it
if you treat it as a single "move" that doesn't go anywhere (begins and ends at the same place) then its simpler
not really, the whole structure of threading isn't like that
sure I understand
that goes down to depth and returns to the same place
but in TP, currently threading needs lines to follow
you mean lathe threading?
I should call it spindle synchronized motion
tapping and lathe threading are completely different things IMO
tapping never has leadin or leadout moves, its always in a straight line
s/threading/spindle synchronized motion/g
and it always reverses and ends up in the same place it started
I saw you say that
but I disagree (I think)
the basic move is the same, you track the spindle proportional to the thread pitch
in the ideal case, yes, it would be nice to be able to track any move or series of moves in either direction
(edm guys would love that)
but the implementation is a nightmare
I'm considering allowing backtracking along one move
tapping is a special case, because the tapping move can be treated alone
it will never be an arc, it will never have blending, etc
and it will always be a single move
(I can't decide which implementation will suck less)
steves_logging is now known as steve_stallings
steve_stallings is now known as steves_logging
[Global Notice] Hi all! As noted earlier, we're about to proceed with a massive upgrade of freenode's server code. This will cause a big outage, but hopefully only for a few seconds. For more information, check the news page: http://freenode.net/news.shtml
let me here a whooo whoot!
[Global Notice] Hi all! All affected servers have been restarted and the network should be returned to normal operation. However, if you notice any problems, please /msg an on-duty staffer in /stats p (or mail email@example.com)
Lerneaen_Hydra_ is now known as Lerneaen_Hydra
jepler: when you're around I would like to talk about debs
jepler: well.. gotta run for a while, but I was wondering if we shouldn't backport the improved dpkg stuff to 2.1
I built and ran packages successfully
if it works, I'm in favor of backporting it too
ok.. I'll look over it some more
steves_logging is now known as steve_stallings
jepler: did you take a peek at the debian stuff I added?
alex_joni: I just saw the names of the added files
those are just copied from the existing packages, right?
the configure change looks right to me
that's what I was wondering, thx
which reminds me - this is the reason I had brought up GPL recently (with Alex, I think)
a week or so ago, I asked a hypothetical "how might the GPL affect me" kind of question
relative to emc/HAL
* alex_joni is getting old
memory is a fog
getting married is the problem ;)
oh, I was hoping it was part of the solution
no definite conclusions from the conversation though
steve_stallings is now known as steves_logging
cradek: I looked over the things jeff did for packaging today, added a few things and built / installed packages for dapper
cradek: I think it works quite OK, and I'm tempted to backport it to 2.1
but I'd rather do that after you look at this too
I trust you... I'm sure the stuff that's currently in 2.1 is not at all ready or tested
it needs a bit of fiddling with depends and all to make it work
the stuff that currently is in 2.1 I mean
the stuff in head makes it easier to build for different platforms
yeah, I think that's great
and it accomodates for the platform extra stuff
I could never decide if that stuff belonged in the emc cvs, but I guess everyone else thinks it does, which is fine by me
it's a place like any other
if it has to change from version to version, then CVS is a good place