jepler: cubic splines are representable as nurbs: here is truetype-tracer output from a postscript type1 font: http://timeguy.com/cradek-files/emc/asdf.ngc
(some computer-modern font picked at random)
cradek: that's good to know
probably the conic splines are too, then? (G5.1 ... L3 ?)
(hm, is L related to the number of points ? Is it necessary to specify it at all, then?
(or else G5.2 is redundant?)
L is the number of control points considered at a time. it's <= the number of points given (+- fenceposts, I dunno)
so no, the end code is needed
cradek: Still around? Yes, I'm bringing the quadrature card in question to the fest (and some things that are quadrature-encoder-like, whatever they turn out to be, LOL.) Also two PCs (the other one with Mesa cards). And I *really* hope a big flat screen, we'll see. And cords, power strips, etc. Tools. Scope? I guess I should put this stuff on the wiki, now that I started a 2010 page.
cradek: Have you heard of anyone making an extended stay (other than Monday for the exhibitors)? Either arriving early or leaving late? Or both?
jeff and I will be in ohio for the preceding weekend - not sure when we will drive home
cradek: I could use some advice about whether to run the emc install script that gets the key, points apt-get to hardy 2.3 (2.4?) etc. John's PC has the 2.4.0 parallel port issue so I backed it to 2.3.5. But I thought 2.4.1pre might help with some other stuff if I can avoid the parallel port issue. I'm trying to do make, --enable-run-in-place, etc, have never done it. Should I run the (2.4 modified) install script?
cradek: I mean, it seems that on one hand I *must* because I want apt-get from the 2.4 sources, but on the other hand I *must not* because I don't want to load anything that will bring back the parallel port issue. So I'm not sure what to do. Any advice appreciated. Thanks.
KimK: you don't want to run the script for 2.4
you still need to have 2.3.x installed (as that one is working ok)
to test 2.4.1~pre, you need to grab a copy from git
it's described in the wiki how to get the code, and compile it in place
did anyone tried to run in 64 bit lately?
aystarik: why, any issues?
latency test prints something looking like max-int
both native rtai and emc
* JT-Dev wonders if my problems trying mozmck 10.04 is related to the video driver it didn't have?
aystarik: hmm.. it worked ok a while ago
aystarik: I think jepler had 64 bit rtai working and it seemed fine except the machine hung occasionally (weekly?)
but yeah, it was not too recently.
it's been literally years.
I run sim on 64-bit machines all the time.
are you using the 8.04 kernel from linuxcnc.org, or a different kernel? that's the kernel that I know was good way back then.
(good except for the eventual lockups, whether I was running rtai or not -- but that PC was always just a bit flaky)
jepler: years, but not more than 2 :P
no, I suppose not
if git clone asks for a password does that mean the .ssh/id_dsa.pub file is bad?
hmm, I have the id_dsa file too with the private key that I copied from another computer that has push access
oh crap I think I figured out what I was doing wrong :/
yep that is it my name is not "developername"
maybe I should read what I copy before I paste it LOL
EMC: 03jepler 07v2.4_branch * r015304ab3bd5 10/src/hal/user_comps/modbus.c: compute correct response size for read_input_registers
EMC: 03jepler 07v2.4_branch * r43c2039a7c03 10/docs/man/man9/encoder.9: fix markup problem
EMC: 03jepler 07v2.4_branch * r416b8337e708 10/docs/src/drivers/pluto_p.lyx: multi-board support for pluto is unlikely
I just installed 64-bit 8.04.4 in qemu, added the linuxcnc.org kernel, and get abysmal latency test numbers (kern/latency and latency-test) but not 'something looking like max-int')
well, I wouldn't expect smaller numbers in qemu
if the numbers are bad but changing, that means it works
looks like I screwed up that change to pluto_p.lyx
mozmck if your around I made it past the rtai in 10.04 finally
10.04 I have mozmck 2.4 installed but it doesn't give me a chance to select the rtai kernal
johnt: hold down the shift key to access the bootloader menu. The Bootloader should not have any visible representation, unless the User holds down the Shift key, which will trigger the text-mode boot selector.
[17:01:50] <jepler> https://wiki.ubuntu.com/DesktopExperienceTeam/KarmicBootExperienceDesignSpec#Bootloader
jepler, thanks I tried every other key I could think of :)
I'm getting somewhere now I'm up to locked memory :) brb after I fix that
I got a new hard drive for 10.04 so I wouldn't have to trash 8.04, but now I've discovered the drive won't work in that computer--an early sata_via chipset that won't talk to 3.0Gbps drives, and the drive has no jumper setting to select 1.5G :(
the machine's not even 5 years old
I think that existing hard drive has 6.06, 5.10 and maybe even a BDI so I should just wipe one of them instead
no bdi .. but it does seem to have windows XP
jepler: 10.04 may resize the partitions and install beside 8.04 without any problem.
it does great doing that with Vista and Windows 7.
for some strange reason I didn't manage to install it on my laptop
had some reserved space free for just this purpose
it did install, but the bootloader was just fscking things up (and I tried both grub2 and grub)
fwiw I successfully resized 8.04 and installed 10.04 alongside it
(insert standard disclaimer about how you'd be insane to try that without backups)
but since you do your regular backups anyways, it's no issue
latency test running on AMD Sempron 3000+ (1.8GHz), 12ns jitter after a minute or so
I don't recall what I got under 8.04 since I use it as a servo machine
no issues besides the known ones (soft memlock was set to 64)
hmph! I started glxgears and got 75us latency, even though glxinfo says it's software rasterizer
(it's not onboard video either .. Matrox G400)
10.04 with mozmck's packages officially moves my little mill
here is where I'm at trying to run 2.4 on 10.04 EMC error http://pastebin.com/M5gy0NS2
trying to run axis sim
the latency test loads up but does not run also
* johnt wanders out to the shop to make a few parts on the Hardinge
you've got a bunch of milltask processes. if you can't kill them in any other way, restart. Other than that I don't spot anything that looks like a real error.
er, no, you don't -- the output of ps is repeated over and over
I can't tell what's actually wrong then
jepler: thanks for looking
Can not find -sec EMC -var NML_FILE -num 1 ........ Isn't this the issue with the NML= _____ file in the EMC section of the ini file.... I've seen this error before ... 2.4 doesn't require it I believe .. put it in and it does this I think
I'm using the 2.4 axis sim... let me run down to the basement and see if it is in there
So comment out your NML= line and restart ..
no NML= line in that file
If you are using 2.4 comment out that line ...
there is no NML= line
In the system that is having this problem?
the other strange thing is the latency test pops up but does not run
yes this is a brand new 10.04 install with 2.4 installed
Well... then you have got me ... perhaps that error shows up due to other things also ..
when I told EMC to add a desktop launcher it won't let me use it as it says "Untrusted application launcher..." Cancel
when I click on it
johnt: try increasing DEBUG and load sim/axis again
I had to go to the permissions tab and check execute
If I put the nml= into the ini file I get a bunch of waiting for s.axes ... also but I also get different NML related message before that ... so that is obviously not it.. sorry ...
EMC errors http://pastebin.com/9iFfQthk
johnt: ini and hal also please
axis.ini sim http://pastebin.com/MwpyJtfN
micges: it's sample configs
johnt: you're using mozmck kernel packages?
alex_joni: I see, but it's strange
asus motherboard with AMD quad core
the problem is motion not coming up
ATI video card
backlash is the first thing that gets sent to the motion controller
I suspect RT stuff is failing
setaxes before failing too
should I reinstall the packages
but I'd like to see a dmesg, but with higher rtapi error levels
micges: setaxes calls set backlash iirc
err.. no, you're right
but both talk to motion in the end
alex_joni: no but it's meainingless
johnt: try this from a terminal: sudo /etc/init.d/realtime start
echo 5 > /proc/rtapi/debug<tab>
I think it's debug_level, but it's been a long while
then try to run emc
nothing happened alex_joni
check dmesg and see if there's more messages now
should I clear dmesg first then run it?
and change DEBUG=1 to 5 or so
then rerun emc2
rtapi.conf is empty
that sounds bogus to me
yes, double checking now
actually there is no /etc/rtapi.conf
maybe /etc/emc2/rtapi.conf ?
got one there
jepler: should we distribute pyc's with the emc2 deb? or regular py's ?
EMC Errors http://pastebin.com/FaH27FB3
insmod: error inserting '/usr/realtime-2.6.32-22-rtai/modules/rtai_hal.ko': -1 File exists
sudo /etc/init.d/realtime stop
johnt: not usefull ;)
there need to be a lot of things in there
at least a couple lines more
(running realtime start first was a bad instruction from me..)
johnt@lucid:~$ sudo /etc/init.d/realtime stop
ERROR: Module hal_lib does not exist in /proc/modules
ERROR: Module rtapi does not exist in /proc/modules
ERROR: Module rtai_math does not exist in /proc/modules
ERROR: Module rtai_sem does not exist in /proc/modules
ERROR: Module rtai_fifos does not exist in /proc/modules
ERROR: Module rtai_sched does not exist in /proc/modules
ERROR: Module rtai_hal does not exist in /proc/modules
then something borked..
you probably will fix it easiest by rebooting
after clean boot EMC Errors http://pastebin.com/gHR8v12X
and dmesg http://pastebin.com/7Jp84DiW
yeah, dmesg was included
hmm.. all looks ok
can you start from a terminal, with DEBUG set really high (0xffffff) and pastebin the whole terminal info?
alex_joni: output seems ok but motion seems not responsive, maybe some shmem problems?
micges, also note that the latency test does not run
alex_joni, much of the terminal info scrolled off the screen
is it possible I might need to reinstall rtai?
johnt: not even rtai's latency test?
from the terminal window? latency-test
not that one
make sure emc2 isn't running when you do that
No such file or directory
there is a testsuite
my memory isn't the youngest anymore
I'm sure I have you beat on loss of memory :P
it starts but no numbers show up for the columns
hmm.. sounds like timer isn't working or something like that
leaving RTAI tasks frozen
johnt: sorry, no idea what to suggest
can you pastebin a comple dmesg after boot?
without clearing it?
or pastebin /var/log/bootlog
or whatever it's called like
bootstrap is the wrong one
boot is empty and boot.log has machine code in it
then only dmesg after a reboot
boot.log has stuff about batteries and sensors etc in it
hi.... i'm learnnig HAL programming, is not difficult
i have a question: what parameter/function capture first the commanded position by gcodes?
bootup dmesg http://pastebin.com/TxGfbKzC
is axis.N.motor-pos-cmd ?
penguin: I don't fully understand, axis.N.motor-pos-cmd is commanded position of axis after interpolation
updated every 1ms
the first first function that catch the position is that?
before send it to stepgen
(i want to know how to catch the gcode commanded position before send it to stepgen)
yes this is position that should be sended to stepgen
johnt: I'd try booting with isolcpus=3 as a bootoption
I think RTAI likes to use the last cpu
check with SWPadnos, he might know isolcpus more than me
alex_joni, where do I put that at?
micges: then, axis.N.motor-pos-cmd "read" gcode positions before anything
that means you hold shift for grub to come up
then you select the line you want to boot (the rtai kernel) and press e for edit
then you can add what I said to the end of the line
penguin: I think you can call it that way
micges: in the same way, the "S" command (pwm value) is captured in first instance by motion.spindle-speed-out ?
didn't make a difference alex_joni
all out of ideas
thanks for trying
micges: so, i can process "S" value before send it to pwmgen (for example)
(make a function that takes S value and output pwngen value)
alex_joni: Hi, thanks for leaving me an answer on my earlier Q. Yes, I have installed git, etc. on that machine, and I was following the "Building EMC realtime" advice toward the end of wiki page http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Rastering_With_A_Laser,
and it started off OK, but when I got to "make" it failed with an error: Makefile:31: Makefile.inc: No such file or directory <newline> Makefile:42: *** Makefile.inc must specify RTPREFIX and othe
r variables. Stop. I've never done the whole make-your-own and --enable-run-in-place and so on, so I don't know what to do next. I would like to be able to run the standard release as usual, and whatever "previews" are available through git. I would appreciate any advice on this error.
Sorry, got long-winded again
did you run autoconf.sh and configure first?
Yes, I was following their list, <cd src> <./autogen.sh> <./configure --enable-run-in-place> [seemed OK so far] <make> [then got the two errors] So I never got to <sudo make setuid>
Then I thought, maybe he wanted to have already run that install script to get some missing files?
he wanted to --> he wanted me to
if configure completes ok, then you have what you need
but I doubt that
try 'sudo apt-get build-dep emc2'
* alex_joni is off to bed
good night all
good night Alex
good night alex
johnt: So latency-test will run but it just shows 0 for everything?
alex_joni: Thanks for your advice, sorry I hung up on you, lol, I certainly didn't mean to do that. I went to try doing what you said on the PC in question, but I tried to do it remotely using "Remote Desktop Viewer". Unfortunately, I forgot that I discovered yesterday that "Remote Desktop Viewer" (Vinagre) has some kind of bug and when I "connect to..." with it, I get the black screen of death, so I have to remember to use KRDC, which seems to work OK. Bu
t I forgot about it today and so I froze, had to restart. Hope you see this later, thanks for your help. I'm trying it now.
ries_ is now known as ries
alex_joni: (when he returns) OK, it compiled for a long time (slow machine) but may be OK? I can't test it right now, machine is in use, but will try it as soon as I can. More news after a while...