patch -p1 blah blah blah ;)
alex_joni: thank you for working on Hardy
jmkasunich: thank me when/if it works :)
whee.. patch applied
dang.. this is quite a fastish machine
the last one was my 1400MHz athlon
yeah, hard to get anything that slow these days
mobile, 1M cache??
ah - 1M per core, dual core?
1.73 GHx, google tells me
yeah, that one
can one of you put a dapper config online?
a dapper .config?
saves me booting another machine up
rtai, or stock?
I think there is one for the 2.6.20 SMP kernell that cradek made
let me boot my RT dapper machine
ah, nm.. I can do the same
(is the file on every box, or only those that built kernels?)
I thought maybe one of you had one running already
jmkasunich: every box
I should leave it on, since its the compile farm RT dapper "slot"
are you goingto try for SMP?
SWPadnos: hmm, not yet
it should be /boot/config.gz or something
I can use the shoptask
I'm booting a VM right now
heh - wiki.linuxcnc.org/uploads/config-2.6.15-magma
unless that's upload with no 's'
[00:09:14] <jmkasunich> http://jmkasunich.com/pics/config-2.6.15-magma
9th June 2006
not bad for a 6.06 release :)
* alex_joni prepares for a bazillion enter's
how many is a bazillion?
I suspect a couple hundred at least
heh - I think the quote is "how many is a brazillion" :)
how many IPIPE domains do we want?
make oldconfig :)
SWPadnos: that's what I'm doing :)
make oldconfig < yes
hmm, doesn't work
there are numeral choices aswell
or true or whatever it is :)
there is a make option actually - make allyesconfig or something. not sure if those can be combined with oldconfig though
allyes, allno, allmod ...
hmm.. I did something else on dapper iirc
I started from the dapper .config, and used that as a basis for oldconfig
then changed the RTAI dependent stuff
are you using some RTAI or have you taken the drastic step of going to xenomai (with the RTAI skin)?
I would have used 3.6, but there are a couple of bugfixes which I definately want
the >1G memory fix is one
in a couple of weeks, I may be able to help out with this - I think SMP is important enough (or will be before 10.4 LTS) that we should put some effort into it
module versioning is another feature, although that might be present in 3.6-rtai
oh, interesting. isn't that one thing that needs to be turned off right now?
it used to
"right now" meaning "in the version that the distributed EMC RT kernel uses" :)
oh, sure.. in dapper it's turned off
preempt the big kernel lock <- wtf?
be vewy vewy careful
BKL shouldn't be used anywhere now anyway
besides the obvious name ..
big kernel lock
oh - way back when, you could basically lock the whole kernel when you were doing something, rather than a semaphore/spinlock on some particular resource
on uniprocessor machines in 1993 it wasn't that bad a decision to make :-P
(particularly annoying on SMP machines, because I believe all CPUs were locked by the BKL)
sounds acceptable for a full build :)
kernel build + creating the deb's
crap, r1 is a no-go :)
51:15 * n
and ccache at n-1 usually
I installed a horny kernel update on a machine today and it porked the radeonfb console
so I booted the previous kernel again. something-12
the working one was something-9 I think
seems the last one is 12.22
12.20 was a bad one :)
yeah, that's what I'm running atm
-7.12 is the one that works - I didn't get any intermediate ones
I also have 11.something installed
arch/x86/kernel/entry_32.S: Assembler messages:
arch/x86/kernel/entry_32.S:270: Error: invalid character '_' in mnemonic
do you think this is the last one?
fun.. ain't it?
pain in the neck
at least it fails pretty early
alex_joni: so are you adding the rtai patch to the ubuntu kernel, or to the vanilla kernel closest to the ubuntu kernel?
trying vanilla so far
but I think I'll try ubuntu next
I should get a 64-bit install of hardy going
I wonder why my first build worked without issues..
and now I always get that asm error
* jepler salivates over what $700 buys: AMD 4 core @ 2.2GHz, 8GB RAM, DVI video, 500GB SATA 3.0Gb/s, nice Antec Sonata III case.
AMD/ATI chipset, so there's even a chance of an OpenGL driver written from vendor-provided datasheets (not holding my breath on that one)
sounds nice :D
yeah, too bad I just spent all my money on a new car
hrmpf.. with a certain config file it compiles..
jepler: noticed the python-tk message earlier?
maybe configure could catch that?
did you get the car today?
how is it?
(if you spent your last $700, you did it wrong)
alex_joni: this command will exit with success if tkinter is installed, or with failure otherwise: python -c 'import _tkinter' 2>/dev/null
it works OK even if DISPLAY is unset or set to something invalid)
jepler: I'll add it to configure tomorrow then :)
alex_joni: IMO it's unimportant for 2.2
for TRUNK it's a fine addition
sure, wasn't thinking about 2.2
cradek: the car's fine, and I didn't really spend my last $700
we haven't driven it much yet
hmm. it seems the rtai patch applies to the hardy kernel too
huh, with highlighting, internal wiki links to attachments are damaged (on the ubuntu wiki): https://wiki.ubuntu.com/HardyHeron/Beta?highlight=(png)
maybe just for attachments that don't exist yet..
weird - I could have sworn that the "sample every N runs of the thread" functionality in halscope worked at one time
but CVS says it has been untouched since 1.1
what is "it"?
the line of code that prevents it from working ;-)
a very simple mistake on the call that creates the spinbutton
you said it
can we argue about the trigger line now? :)
the upper limit was set to 1
(I hope the rest of the code works ;-)
seems the paravirtualisations config is the one borking things up..
when that is unset, it seems to compile gladly
EMC: 03jmkasunich 07TRUNK * 10emc2/src/hal/utils/scope_horiz.c: fix 'sample every Nth time' function of halscope
* alex_joni calls it a night..
see you guys tomorrow
after all it's 10PM here
(maybe my second kernel will boot.. the first one didn't)
wait - are the Nebraska guys calling it a night at the same time as the guy who is about 8 times zones ahead?
not me (yet)
jmkasunich: it's roughly 5am here :/
oh, I sleep late :)
probably I'll be up for lunch around 2pm :D
EMC: 03jmkasunich 07v2_2_branch * 10emc2/debian/changelog: backport 'every Nth time' fix for halscope
EMC: 03jmkasunich 07v2_2_branch * 10emc2/src/hal/utils/scope_horiz.c: backport 'every Nth time' fix for halscope
EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/spy_vars_gtk.c: chanages so var window changes symbol names when section window button toggles plus more...
EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/zEMC_README.txt: update change log
morning.. anyone around?
hi, just got up
hmm.. I have 2 news re hardy
managed to boot a pached kernel
(some stuff still missing, I noticed they don't include modules in the image deb anymore, so I need to figgure out how to build a lum deb)
the bad part is that rtai doesn't want to compile using the freshly installed headers :/
can anything else use those headers to build kernel modules?
hmm.. I could try emc2
the build starts, but then fails while looking for irq_vectors.h
which is present in include/asm-x86/mach-default
found some old references on RTAI mailing lists, that that location should be picked up by the build system
do you think it's a problem with the includes themselves or the header deb packaging?
hmm.. I tried with the sources dir, same problem
so I don't think it's the header deb
what can I do to help?
looks like 2.6.20 is the latest kernel I've used
not much atm ;)
trying rtai-3.6 now
(maybe you can find some info on building a l-u-m deb on the ubuntu wiki somewhere.. what I found so far is pretty scarse)
ok, I'm off to shower and breakfast
see you later
maybe I should install hardy somewhere
doesn't make-kpkg make that?
not with the default kernel_image/headers/source incantation
[13:45:52] <cradek> http://ubuntuforums.org/archive/index.php/t-619171.html
anything else for 2.2.2?
EMC: 03jepler 07v2_2_branch * 10emc2/debian/changelog: univstep config fix from TRUNK
hmm.. rtai-3.6 seems to go way better than vulcano
EMC: 03jepler 07v2_2_branch * 10emc2/configs/univstep/univstep_load.hal: univstep config fix from TRUNK
(well, besides that)
jepler_: did you see the pluto patch?
wonder if that's too much of a version name :P
cradek: thanks, got it now
if only I was set up to test that hardware
EMC: 03jepler 07TRUNK * 10emc2/src/hal/drivers/pluto_step.comp: from Len Shelton: fix negative scale values
EMC: 03jepler 07v2_2_branch * 10emc2/debian/changelog: from TRUNK: Len Shelton's fix for negative scales
EMC: 03jepler 07v2_2_branch * 10emc2/src/hal/drivers/pluto_step.comp: from TRUNK: Len Shelton's fix for negative scales
er, oops -- wrong name
* jepler_ really crossed his wires on that one
bet cradek can fix the cvs comment
any of us can use 'cvs admin' to do that
hmm.. rtai-3.6 compiles against the source, but not against the headers dir
cvs admin -m1.11:"from Michael Jochum: fix negative scale values" pluto_step.comp
cradek: bad headers package?
missing a symlink or some files somewhere?
it had a borked symlink
I changed it to asm->asm-x86 as it should be, but still nothing
ah, found it
* jepler_ twiddles his thumbs waiting for the hardy x64 installer to start
what the HECK is this bird background in the installer?
weird weird weird
I guess it's a heron
jepler: same issue? http://cnczone.com/forums/showthread.php?t=54585
oops - sorry - some reason I thought it was talking about pluto step. (campfire last night... lots o 'fun)
I don't know what the problem is with pwm+direction in pluto-servo, the other problem was something to do with pluto-step
EMC: 03jepler 07TRUNK * 10emc2/src/hal/drivers/pluto_servo.comp: another typo was keeping pwmdir from working properly
I had written 'i' instead of 'x', and wasn't able to see it
EMC: 03jepler 07v2_2_branch * 10emc2/debian/changelog: fix pwm+direction mode
EMC: 03jepler 07v2_2_branch * 10emc2/src/hal/drivers/pluto_servo.comp: fix pwm+direction mode
I'll wait anothre hour in case there's another bug fixed :-P
will this be backported?
I see some things in stepconf
(working on one fix, the other I want to talk to you about)
EMC: 03jmkasunich 07TRUNK * 10emc2/src/emc/usr_intf/stepconf/stepconf.py: improve calculation of HOME_LATCH_VEL
jmkasunich: the fix you just checked in looks good, and it looks like a bugfix so I'd like it backported
I'm doing that now
* jepler_ twiddles his thumbs waiting for the first hardy 'apt-get upgrade' to finish
EMC: 03jmkasunich 07v2_2_branch * 10emc2/debian/changelog: improve calculation of HOME_LATCH_VEL
EMC: 03jmkasunich 07v2_2_branch * 10emc2/src/emc/usr_intf/stepconf/stepconf.py: improve calculation of HOME_LATCH_VEL
the other thing about stepconf is defaults, and maybe its no big deal, since people should change that stuff anyway
but some defaults seem out in left field
in mm mode, the default screw pitch is 20mm/rev
and default max velocity is 1mm/sec, which is very slow
default home search vel is 0.05 mm/sec, which is glacial
ah, I see - the defaults are the same regardless of mm or inch
changing that would be more complex, and is not really a bugfix
I just noticed something weird (but harmless)
when you first enter an axis screen (say X), all the values are displayed as integers "1" "30"
if you proceed to the next screen, then back to the one you were at, the values are now floats: "1.0" "30.0"
2.2.4 hapening this weekend?
I believe so
[17:54:16] <skunkworks> http://cnczone.com/forums/showthread.php?p=425929#post425929
jmkasunich: yeah the system of defaults stinks
jmkasunich: but I don't see a simple solution
it can wait for 2.3
anything else last-minute for 2.2.4?
lemme check the trackers
(not that we're likely to see bugs with 5 minute fixes in there)
did you fix "pluto-step doesn't handle negative scale values" ?
yes, I blindly applied the patch
I guess we can close that one, if the patch doesn't work somebody will complain
whoops, I thought I had
I don't see any other "quick-n-easy" ones there
EMC: 03jepler 07v2_2_branch * 10emc2/debian/changelog: note the bug number that this fixed
I think I will go ahead and start the release process now
EMC: 03jepler 07v2_2_branch * 10emc2/VERSION: bump version for release
EMC: 03jepler 07v2_2_branch * 10emc2/debian/changelog: bump version for release
wow, "cvs tag" sure isn't fast
EMC: 03jepler 07v2_2_branch * 10emc2/VERSION: bump version after tagging release
EMC: 03jepler 07v2_2_branch * 10emc2/debian/changelog: bump version after tagging release
jepler: I'm around
cradek: great -- I will let you know when the updated packages are on dreamhost
(Locally modified changelog moved to .#changelog.184.108.40.206)
dear chris: I need to develop some cad software. can you tell me how?
cvs update: warning: debian/changelog was lost
cvs update: move away `debian/changelog'; it is in the way
here's the thing that often happens to me when I 'cvs up -C': ^^^
I never do cvs up -C, I cheat and delete any modified files, then do a regular up to replace them from the repository
I often do cvs up -C, but I don't think I've seen it cause that
cradek: was that a mail you got? ("teach me how to program")
that was a pretty fair paraphrase, yes
jepler_: did you regenreate configure?
cradek: not in CVS
I hate anything that says "I _need_ to do foo" - that means either "my boss told me to do it" or "my teacher told me to do it"
jepler_: ok just checking
jmkasunich: in this case, I think it's boss
"I want to do foo" is much more promising
jmkasunich: no sign that he's developing Free software either
cradek: my procedure runs autoconf before it runs configure, but that regenerated configure never gets put in cvs
it seems to have something to do with the file being a conflict .. http://pastebin.ca/945118
after the 'cvs up -C' completes, the halrun.in file still contains the same data as it did before
it must be deliberate
jepler_ is now known as jepler
EMC: 03jepler 07v2_2_branch * 10emc2/src/configure: bump version after tagging release
packages uploading now, but it'll be 10+ minutes before they're done
oops -- still more packages to transfer
starting on the rt ones
cradek: about 3 minutes left
cradek: I'm ready for you to do your thing
cradek: had a chance to sign the repository yet?
jepler: yay on the release
cradek: I think there's a bug in make-kpkg (it sets a symlink wrong)
I'll try to build debs with the ubuntism hacks ..
alex_joni: it's not really released until chris signs it, and I don't think he has yet
(it won't appear in apt-get update / update manager yet)
jepler: yeah, I know.. I can still be glad it happens
oops sorry, I was away
The following packages will be upgraded:
I'll work on the wiki shortly
jepler: I'll do SF and linuxcnc.org in a bit
Installing new version of config file /etc/emc2/Makefile.modinc ...
wonder what the heck that is :-)
wasn't me :P
thanks for doing the release jeff
what a nice list of fixes
quite a lot of them :)
best emc release ever
but, I always say that
(I believe it too)
well.. it might always be true
classicladder/calc.c is userspace?
it looks like it includes RT headers ...
jepler: I notice the 2.2.4 tar.gz is about 90kB smaller than 2.2.3
alex_joni: huh, I don't know why that would be.
jepler: I downloaded it from cvs.linuxcnc.org.. not sure either why
(but I noticed after doing the SF release..)
btw, SF and linuxcnc done
I did modify some of the .png images to change the embedded dpi setting; I wonder if gimp compressed them more effectively than whatever had been used before
alex_joni: they're about the same size on www.linuxcnc.org
5844 -rw-r--r-- 1 emcboard pg167418 5975748 2008-01-27 18:37 emc2_2.2.3.tar.gz
5844 -rw-r--r-- 1 emcboard pg167418 5974573 2008-03-16 10:30 emc2_2.2.4.tar.gz
emc2.2.4.tar.gz : 5890306
emc2.2.3.tar.gz : 5975748
less bugs in the new one (you didn't realise bugs took up so much space did you?)
* alex_joni calls it a night
good night all