I don't really understand why host and client couldn't be more compatible with each other
oops - gotta run. Hous of Prime Rib time!!
wooo found scanline combination that worked with crappy monitors
is ubuntu blackbox broken or something?
okay i figured it out it wants the menuFile line at the beginning of .blackboxrc
new blackbox is kinda lame...
03cradek 07TRUNK * 10emc2/src/emc/task/emccanon.cc:
document petev's question about arc parameters, fix a wrong comment,
and undo some `indent' screwage
lerman_ is now known as lerman
SWPlinux: re petev's suggestion... right on, i had thought about using .v (verilog) files in the same way. the advantage to either is: you put stuff where you want stuff and the 'wiring' follows, yet can be modified. either format can be parsed. good thinking petev!
what suggestion is that?
somebody suggested using geda for HAL stuff earlier today, was that pete?
jmkasunich: thats what i read, installing gschema and PCB now to look at the idea
Its something thought about probably a year and a half ago
just didn't have the time to run with it
oops, dont use synaptic to install gschem, you get only gschem :( with no manual :( instead use 'sudo apt-get install geda'
found symbol creation guide online http://www.vision.caltech.edu/bouguetj/ICCV98/index.html
eric is not in the sudoers file. This incident will be reported.
will I have to go to the principal's office
it really does mail root...
kind of silly when it's so easy to do that on the wrong machine
what happens if root never reads his mail?
it builds up forever?
it will eventually fill the universe
can you delete it without reading it?
I've got some robots that have been getting mail from the cron jobs that run automatically for years now
no yum package for geda apparently
how did crepinc do a test in between the time that he left and entered the room? spooky
there is more than one of him
see the ` at the end of his name
* ejholmgren smacks root with a rolled up newspaper
Stop smacking Mr Robert Oot!
built 1st component in gschema and saved to a library.
some local dot files (~/gEDA/gafrc) have to be manually created.
Symbol files have to be sudo cp'd into /usr/share/gEDA/sym/local.
There will be some discussion about appearances ( IEEE symbols dont seem applicable )
and elements like gain become pins connected to constants.
still lookin at all this, it's just like new components in CircuitMaker et al.
[04:30:12] <tomp> http://imagebin.org/8012
tomp, that doesn't look half bad
which half :)
if you need some help with a netlist parser, let me know, I can probably make something up pretty quickly with ANTLR
the other half ;)
SWPLinux, how does it feel to be on PST ?
I looked at the text representation of a component (in the help files), and it doesn't look too nice
SWPLinux: when are you learning?
petev: not bad, I usually keep west coast hours even when I'm on the east coast :)
I'm learning all the time
I leave tomorrow morning
packing as we speak
oh nice short shuttle ride
no shutt;e from the Hilton, strangely - taxi for me
you can take the light rail if you want it cheap
nah - I'd rather get the extra sleep
taxis are fast and on demand
and pretty cheap, unless your cell phone falls off in the back
cheap?! is probally a $30-$40 ride
$15 with tip
at least, that's what it was on the way here
didn't think it is that close...
3 or 4 miles, I think
it is a 15min light rail ride....
6 or 8 in a cab
so you won any raffles this year?
[05:12:11] <SWPLinux> http://www.google.com/maps?q=300+S+Almaden+Blvd,+San+Jose,+California+95110,+USA&sa=X&oi=map&ct=title
nope - didn't really enter any (or stick around for them)
I did well enough last year, so I still average out OK
so the paid talks don't have give away raffles?
(512M ipod shuffle, 1G secure jumpdrive, 128M i-stick ...)
no, but I did get permission to do limited distribution of a few of the papers
not that that's really a prize or anything
did you goto any of the prius stuff?
no, I missed all that
actually, the 3 talks I went to today were excellent
all by the same guy - one on memory-mapped I/O (and other things) in C++, one on C++ templates, and another on the behind-the-scenes stuff C++ does for things like virtual functions and stuff
got your money's worth?
it's surprising, but in many cases, using templates (which people including me) think are the worst thing for bloat, are actually better sizewise than C
yes, I think so
sure sounds like the paid track is very different from the free stuff
there aren't enough advanced or "good intermediate" courses for my tastes, but there are always subjects where an introductory course is enough (VHDL, Verilog, control theory ...)
you can get credit through UCSD for the paid stuff
but what's the credit good for?
maybe I'll get a degree one day
well, degrees ...
really? as in a MS?
at least BS, and maybe MS - I didn't attend that session, unfortunately ;)
I can check it online, but I suspect that enough of the advanced courses might be good for an MS
they just started that program this year
might do a paid track next year then
You do VHDL/Verilog?
I usually scan the brochure and make sure that there are interesting things for around 1/2 the time slots or more
yes, intro to each of them
what tools do you use?
the verilog guy has a $12 intro book you can get on his website - he was pretty good
I have Altium Designer
so I use that plus the vendor tools for place+route
ouch... that's a powerful vacuum on the wallet!
want to find someone that has used Icarus to see how good is it
they seem to have a good philosophy though - I've gotten 7 updates (another is on the way in the next couple of months), and they don't charge until they do a new release (usually every 3 years or so)
they've added features, improved compatibility with old file formats (!), fixed bugs ...
the next release will also have their C-to-FPGA compiler
but is it reliable? their old Protel stuff was very crash prone
(at no extr charge ...)
I had Protel PCB 2.78 and Schematic 3.0x (yes, pre-99se)
which I can now load files from, I'm very happy to say
one version of schematic was very crash prone
it seems stable, but like all modern programs takes a long time to start up
oh hehe and not very Linux friendly either
no, but it does work fine on VMWare
checking your license
including the speed/performance aspects?
and every chance I get, I ask for a Linux version :)
they pretty much chewed me out at the booth when I asked about a Linux version
speed is of course slower than native, but I don't know by how much because my main Windows machine is an Athlon 1800 and my Linux machine with VMWare is a dual Opteron 275 :)
I don't really see a slowdown ...
ok, the component pins in gEDA must be drawn in a certain direction ( else all the wires fall off when the component is moved ) http://imagebin.org/8013
that seems a bit funny, you mean it can't figure out to put the red connection point away from the component?
they are very explicit in the manual that the pin ends (the live ends) MUST!!! be on a 0.1 inch grid
yep all snapped to grid, but its the direction of the drag that make the little red end in or out
grid = 100
look at the pics
oh right - I was looking at the tet representation, to see if there might be some way of autogenerating components at build time
it puts the connection end on the component if you draw it the wrong way
that's just lame
it seems to put the hotspot at the start of the line (at least I think that's what it was doing when I tried it out)
i wont do that wrong again , fortunately i could rotate and edit them then sudo cp them again...
now i gotta hand sketch all the comps we have
mi = move flip (mirror)
yep, using the key cmds is the way to use it
that's why I was thinking of autogenerating them ;)
autogen from what input? the .comp files?
not all have .comp files
some are C
from pin parsing in the C files (possibly), and automatically from comp
you might be able to load them and use halcmd api orsomething
hmm. halcmd save symbol ...
I didn't spend enough time trying to figure out the format, and I'm not sure quite where it's documented though (if it is)
is gEDA really useable yet?
all we need it to do is draw a schematic and generate a netlist
it doesn't look it to me, but then again ...
we need the components to put on the schematic first
think gEDA integrates with SPICE, right?
I think the netlist format is OK (if it's anything like the old text ones for Tango or Protel)
the component format is also text, so we should be able to generate components if we want
does it only have one netlist format?
I played with it for about 20 minutes earlier today ;)
seems like eagle + LTspice is the best option AtM
I was going to try to make a symbol for a HAL driver, and realized that the number of pins/params was larger than I wanted to deal with manually :)
why do we need spice?
Altium does OK for that
we want to use it for graphical hal config
simulate the machine ;)
just geta netlist from Mariss or Copley or whoever, and go
spice is nice for analog stuff
it's a simulator
it does mixed signal with digital as well
simulation is a good thing
is the free eagle un-restricted enough to make a HAL config?
now if there is a usable clone of ProE; we should be able to simulate every aspect of a machine cheaply
i dont see any difference in pins and params when in gEDA, they both are pins, is that a problem? is pintype too limited? ( no numeric floats in electronics )
petev: probably not, the sie limitation of 4x6 inches would probably kill you (though I'm not sure what the component limitations are in schematic, that's PCB ...)
if you can fit it in one page
tomp: I thought about that, and don't have a solution
I guess we can't get full type checking, but maybe we can do some
what pin types does gEDA have?
All pins should also have pintype=value attached to them. This attribute describes the kind of a pin. Possible values are: in, out, io, oc, oe, pas, tp, tri, clk, pwr. Please see the Master Attribute Document for more info (looking )
so maybe we can use some of those to get some type checking in the schematic
HAL will catch the rest when you load the config, same as now
yeah - only drive types, not data types
what was the compliment program to PCB ( the tcl/tk app ) ?
it's schematic capture counterpart?
no, it was also tickle
never mind. it uses same symbol files... so same restrictions
with embeded python interpreter , hmmm
well guys, time for me to run. see you on the weekend (unless I have to do honey-do stuff the whole time :) )
have fun, thanks!
thanks, I will :)
this weekend is mountain bike and mill time
have a good flight
thanks - hav efun on the mountain (bike) :)
xcircuits 'subcircuit' feature is like meta-widgets ( make a collection of many hal components into a reusable component, yet keep the original primitives in tact )
you mean like a heirarchical schematic library component?
xcircuit has a user interface that must be blender's great grandfather ;) (2weird4words)
xcircuit is touchy about numlock. the menus work work at all if it's on!
* lerneaen_hydra_ gasp
[12:00:19] <lerneaen_hydra_> http://lerneaenhydra.shacknet.nu/images/files_temp/img_6706.jpg
<-- any idea why when I cut small threads the profile is all fscked up
it's not 30 degrees from the normal
more like 40/20
maybe the CAM output is all fscke up?
no, the code is correct, at least I think so
same Z values for the thread
have you ever got threads like that?
nope.. I never threaded though
can you pastebin the code?
[12:05:50] <lerneaen_hydra_> http://pastebin.ca/426784
the Z coord is different
the reason for that though is that the lead in is angled
so you have a screenshot from AXIS?
* alex_joni is at work..
[12:09:35] <lerneaen_hydra_> http://lerneaenhydra.shacknet.nu/images/files_temp/temp1.png
I somehow suspect the output from CAM is not right..
but you'd have to ask cradek later
what do you mean the profile?
you mean it doesn't look like a V?
cradek: the sides don't have the same angle
some of that is normal - you'll only see the V if you are seeing along a diameter
but maybe it's worse if the tool isn't on center (height)
cradek: seen the pic LH posted?
I can't really etell if it's bogus or not
* alex_joni cusses at SF
my mail takes ages to get through
cradek: they're not V even when looking straight on
lerneaen_hydra_: ok but there must be a V somewhere where your tool can sit? above or below centerline maybe?
I'm not following you
the tool is centered correctly
poke at it with the tip of the tool until you find the V?
I mean the tool can only cut a V, so the V must be in the wrong place
hard to explain
how carefully did you set the height of the tool?
to within +-0.05mm
I might be wrong but I don't think this is a gcode problem
the nub in the middle after planing is very, very minimal
did you get a lot of lifting of the work while threading?
no, none that I could see
how many passes was this?
I've actually had this problem often, it's most visible when cutting smaller threads
5-6 or so
a 4-40 thread
in brass, with a very sharp HSS tool
I don't see the passes - can you?
err, how would you be able to see the passes?
I mean in the work
oh, wait, what's a pass, the number of cuts needed to make the thread?
5-6 or so
if the cam is wrong, you might see 5-6 steps on one side
I think I had 5 cuts and a spring cut
if you can't see the steps, the final cut touched on both sides and HAS to be a V
oh, the tool doesn't have flank-feed
it feeds straight down
yeah, that's exactly why I'm so confused
then it's a problem with your tool
have to get ready for work - bbl
hmm, I've had this issue with both this tool and standard threading carbide cutters
as well as triangular cutters
lerneaen_hydra_: wish I could see/touch it
lerneaen_hydra_: just for fun can you give us the program or an AXIS plot of it?
cradek: they are above
[13:14:55] <alex_joni> http://pastebin.ca/426784
* alex_joni heads home
cradek: it may be a non-issue, depending on how you tilt the screw the artifacts go from non-existant to awful
lerneaen_hydra_: wait that's where I started! (it's normal)
does a nut screw on?
lerneaen_hydra_: is your cam supposed to cut subsequent passes straight in, or offset along the thread in Z?
uh, it cuts equally much on both sides of the cutter
ie. the G33 move starts at the same Z coord
different X values though of course
there is a python tool that converts structured text into gEDA symbols... http://www.geda.seul.org/docs/current/tutorials/tragesym/tragesym.html
lerneaen_hydra_: I think the gcode is wrong for that - the leadins would all have to be the same length for that to work, since spindle synchronization is at the start of the leadin
lerneaen_hydra_: but since the syncs all happen at the same place, later passes will be at different Z since they have to travel further to get to cutting depth
lerneaen_hydra_: oh maybe not! the Ks change
cradek: I think that the gcode is correct, if not somewhat ugly
I'd have to figure out the length of those leadin segments and compare it to K
too much trouble :-)
I did test a different version however
with straight lead-ins
same result, which leads me to believe that the thread is actually correct
for G76 I made sure the lengths of leadins were the same - changing K is an interesting strategy I didn't think of
IMO a lead-in of constant length is better
as if the total depth is very large then the change in K could take a while if accel is low
then again, there's a change-of-dir accel too, but that won't affect the resulting pitch at least
threading without infinite acceleration is a juggling of compromises (like everything I guess)
lerneaen_hydra_ is now known as lerneaen_hydra
mschuhmacher_ is now known as mschuhmacher
autogenerated hal component using gEDA and this howto http://www.geda.seul.org/docs/current/tutorials/tragesym/tragesym.html
src textfile is http://pastebin.ca/427115
result looks like http://pastebin.ca/427115
isnt really automatic, i still had to hand edit somethings. after a few of these, I'll decide if its just easier to hand edit a text file than to use the intermediate spreadsheet.
[17:20:21] <tomp> http://imagebin.org/8026
Dallur: Are you around?
[19:21:51] <Jymmm> http://www.infoworld.com/article/07/04/06/HNjudgebarsvonage_1.html
that's how much weed alex is going to use this easter
alex_joni he said weed, not canabis
is there a difference/
Yes, one you put weed killer on, the other you smoke =)
never tried any :P
Weed Killer is good stuff, you should try it sometime.
lerman_ is now known as lerman
lerman_ is now known as lerman
fucking bullshit... I'm testing a new AV Sw, made a EICAR file, but xp is refusing access to it with the onyl dialog msg being "Access to C:\eicar.com was denied"