regarding test suite on the farm, I have a couple reservations
1) I'm currently not doing RIP builds
2) I'm not doing make install (which would require me to set up passwordless sudo, and I'm not gonna do that
you can't "sudo make setuid" either, then
2a) even if I was doing RIP builds, I'm not doing sudo ma.... what he said
sim doesn't require "make setuid"
but that won't change the testsuite not working
the question is why am I not doing RIP
(I have no clue, unless we figured the devs are more likely to test RIP so the farm should test normal)
there are (or can be) separate build scripts on a per system and per version basis
so the sim system could do RIP and run tests, while the others just build
that sounds easy enough
my other concern is that some of the failures have resulted in lockups
maybe the 6.06 and 8.04/rt slots should test the full debian package building (that doesn't require root either)
yes lockups will be a bigger risk
I suspect the risk of a hard lock is lower on sim than on rt
it won't kill linux but the testsuite might hang forever
but even a "soft lock" would require either some tricky timeout/kill stuff, or manual intervention
if it becomes a problem I can look at adding a timeout to the runtests script
I have some timeout stuff in the main farm scripts, to deal with cases like CVS down, email down, sftp down, and other such hiccups
but nothing for the build itself
regarding debian, if you tell me what needs done, I don't see why I can't do it
note that we don't have slots for 8.04 yet (rt or sim)
crap, I just realized how old the farm is
the VMs are running breezy (rt and sim)
for a full build I think you would: debian/configure <kernelversion-or-"sim">; dpkg-buildpackage -rfakeroot -uc -us
the only dapper slot is not a VM, and I don't keep it running all the time
(oh, this will require that you put all the stuff for building pdfs on those vms)
I think I should take this system down this weekend and add another 2G of ram, then add some more VMs
for an incremental build, I this captures the intent: make -C src; debian/rules binary
dapper sim and RT, hardy sim and RT
this will have the side effect of dumping debian packages in the parent directory (mumble.tgz, mumble.dsc, mumble.i386.deb and maybe something else)
does that mean at release time, the release manager simply has to copy them?
it is possible for the packages themselves to be signed with gpg, and it's a requirement (or all users get scary errors) that the repostory be signed -- so like passwordless sudo it's something I'm reluctant to suggest
in some ways it would be nice if adventurous users had an easier way to get prereleases, particularly from the stable tree
but that's another separate step from this -- if we want to go this way it *would* get us closer though
turns out I already have a separate script for the -sim slot
[02:25:15] <jmkasunich> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/infrastructure/farm-scripts/emc2_build_sim?rev=1.3
I wonder if we want to build docs too?
if you build the package that will be testing the pdf doc building process
I think I may have disabled that, either because of dependencies that I didn't feel like installing, or because it was taking a long time
both the html and pdf docs are regularly built on a system of mine, but it's not even an ubuntu system
or both :-P
until I have VM based dapper and hardy systems, I think all this is moot
the dapper-rt system is my former dev box - it tends to be off for months, then on for a month or so because I feel bad about the dapper farm being so out of date, then off again because it is a lot noisier than the new system
its been off for a few weeks ;-)
ok, let me know when you are to the point where you need my expertise on the build system
I should be around this weekend
I'll try to get more memory tomorrow evening, and set up VMs over the weekend
I was about to start a hardy download, but... should I be downloading stock hardy and doing the install.sh route, or should I install our iso?
(for the RT VM, the -sim one will use stock hardy)
I'd go with the linuxcnc live cd
* jmkasunich downloads
jepler: has the dapper iso changed since 10/27/2007?
(our dapper iso)
oh wow, I not only have that iso, I have the hardy one too
I guess I can md5sum them and see
-rw-rw-r-- 1 emcboard pg167418 706035712 2007-12-04 12:43 emc2.2.2-1-ubuntu6.06-desktop-i386.iso
looking at the file size is probably enough
-rw-r--r-- 1 juve pg167418 728793088 2008-04-29 16:24 hardy/ubuntu-8.04-desktop-emc2-aj07-i386.iso
even if it's a bit older they're probably similar in terms of number of packages to download/update
yeah, but I might as well use the one we are pushing to users
the 8.04 I have here matches the md5sum of the one online
I'll dl the newer 6.06 one while I sleep, along with stock 8.04 from ubuntu
for one install there is no disadvantage to using the cd you already have and then running updates
the number of updates is about the same?
(note that downloading an iso happens while I sleep a day or two before I do the installs, the updates probably happen while I'm waiting impatiently
ls: .: Permission denied
hahaahahahah that's exactly what I was hoping would happen
I wonder if the 8.04 LTS server edition would be a nice base for a somewhat smaller emc distro
add basic X and a light window manager
reading back, is there a difference between "normal" and RIP builds (when you don't install)?
ok, so the idea is that there shouldn't be, and testing that is a good idea?
oh, there are differences? do you know them off the top of your head?
or some anyway
some paths in scripts, defines in some source files
that's why rip is a build option
oh right, something like module-helper will change
I love fast internet
3 downloads, all sustaining 200+ Kbytes/sec
you have cable?
oh, fast DSL then, 6Mbit?
if only something faster than 1.5 were available here
the one from DH is doing 200K, the two from ubuntu.com are doing 261 and 289
you know, I didn't even have to click to know what that was about ;)
too bad you can't search - surely there's one about broadband speed
goodnight guys, time to walk le mutt
yesterday someone ask for hal designer for emc..
now the best tool for this is Eagle 4.1.6 with eagle2hal packed installed
Designing in Eagle can be done with free version of it
It is very good and VERY usable
lerman______ is now known as lerman
when I enable override limit in axis, I can enable motion even if axis.n.neg-lim-sw-in is True ?
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/config/ini_config.lyx: add filter section
micges: yes that is how it works
motion is allowed in both directions ?
no, only in the positive direction in your example
there are cases where both directions are allowed - consider that neg-lim-sw-in and pos-lim-sw-in are both True (if the axis has only one limit switch and two cams)
in this case emc has to allow both directions and it is up to the user to go the right way
I just wonder of this case, thanks
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (7 files): minor fix to remove error on fresh Hardy build
EMC: 03jepler 07TRUNK * 10emc2/docs/man/man9/motion.9: remove bogus text
I have this situation:
I've stopped machine (table mill) with 2mm diameter in spindle
rpm was 14000
and after stop and moving z axis up there was a wider row in that point
precision was 0.05 mm
what do you mean wider row?
row was 1mm wide
in point when I stopped was 1.1 mm wide after spindle stopping rotating in that point
steppers or servos?
when emc goes into "estop" or "machine off" state, it turns off the servo amplifiers (axis.#.amp-enable-out goes false). it doesn't try to decelerate according to inifile settings or the like. there's no guarantee it will be keeping the commanded position. cutting forces could push the tool around a bit before the spindle stops. if moving at an angle, movement along each axis will stop at a different rate.
0.1mm is only 2 steps
uh, wait. 2mm tool, 1.1mm hole?
I'm with fenn - are you describing something other than a regular old dwell mark that happens when you stop motion in the work?
SWPadnos: tricky, eh?
SWPadnos: this is only example to show what is happening
cradek: I think this is it
I do not understand the problem you are trying to explain with this example then
if it is a dwell mark, this is a machining question, not an emc question; the answer is remove the tool from the work before stopping it
building docs is giving an error: IOError: [Errno 2] No such file or directory: '../docs/src/common/User_Concepts.lyx'
BigJohnT: I think I saw that you added an include of this file .. did you miss adding the file itself?
jepler: yes, and no :)
I forgot I put it in the master this morning and need to add the file now
gee you catch everything :)
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/common/User_Concepts.lyx: note: don't commit before your first cup of coffee or you will forget something
jepler: wooo wooo! chugga chugga chugga
alex_joni: nobody is making text editors to embed in Tk apps these days. you're stuck with mediocre text editors that are probably about 8 years old based on Tk's text widget
jmkasunich: i made an ubuntu 8.04 farm slave vm
i wanna check some things with you before i try it
seb_kuzminsky: did you find the farm scripts in CVS?
jmk pointed me to these instructions: http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/infrastructure/farm-scripts/README?rev=1.6
alex_joni: i havent tested your latest pid fix on my system yet but it seems good
SWPadnos: that was close to poetic :P
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (Submakefile docs.xml index.tmpl): *** empty log message ***
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/config/stepconf.lyx: clear up what to type for a latency test
* BigJohnT kicks cia-37
* BigJohnT don't know what that does but it is fun
EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/config/ (ini_config.lyx stepconf.lyx): clear up what to type for a latency test
EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/common/User_Concepts.lyx: add file
EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/Master_User.lyx: added iportant user concepts to user manual
crap I can't even spell tonight