also I think that we can delete /configs/halui_halvcp
jepler: I thought I removed that :/
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/common/ (Linux_FAQ.lyx User_Concepts.lyx): add info on sudo and change discription of gouging----------------------------
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (docs.xml index.tmpl): remove vcp from html
isn't that ok?
actually I've been talking to chris -- we're at a local pizza place with wifi having "geek night" with another friend
he's working on cutter comp -- there's one last bug
(the one bigjohnt and I talked about yesterday)
I am in my man room ;)
I am breathing a small sigh of relief
I crunched a tail-light on my truck this evening (turning around in a cramped driveway at a friends house, a small tree jumped out in back of the truck)
I was afraid it would be a $$$ dealer part
turned out to be $35
skunkworks: sorry, the phrase "man room" makes me snicker
that's like 1/10 the cost of a dealer part
nice - was that dealer or aftermarket
never bought anything from autopartswarehouse.com before
I haven't bought it yet
jepler: I hoped it would. ;)
its only been broken for 30 mins or so
get right and left, in case they don't match
still 1/5 the cost of a dealer part :)
I'm cheap enough that I'll buy one and see how they match
probably cheap enough to not care unless it's a hot pink vs deep red thing :)
[02:37:21] <jmkasunich> http://www.autopartswarehouse.com/search/?N=10110+1607+4294965327+9252
I suspect it will be pretty decent
hard to tell if it's just the lens or if it includes the reflector/bulb holder part(s)
dunno if you need all of it or just the lens
it includes the bulb holder, I can see thru the lens in the pic
me too, but I don't know if that's a photo of the part "as you'd see it on the truck"
maybe not the actual bulb socket, but I don't need that
[02:39:38] <skunkworks> http://imagebin.ca/img/reXh6w.jpg
notice the one glass
they say "tail light", not "tail light lense"
hey, that's like mine except your wires are like my stacks of paperwork and other junk
that is true
though I've seen "tail light assembly" also
EMC: 03cradek 07TRUNK * 10emc2/src/emc/rs274ngc/interp_convert.cc: fix entry moves that make a very sharp corner, thanks BJT
EMC: 03cradek 07TRUNK * 10emc2/tests/interp/crazy-paths/expected: fix entry moves that make a very sharp corner, thanks BJT
bench is like mine, only I wish I had that much room to mess up
that's the problem with a ~180 square foot office
takes longer to reach critical mass tho
well, I've been working on it for 6+ years :)
this is actuall after cleaning a few weeks ago.. It doesn't take long
wanna come over and be my friend?
jmkasunich: how is the robot?
I hope to mill the two plates that make up the main body today, if I do we should be able to do most of the assembly tomrrow
hmm - my radioshack variable powersupply doesn't keep voltage very well. I guess I need to take it apart
I thought the scope was screwing up.. until I looked over and saw the meter changing
heh - stupid trimmer is dirty
EMC: 03cradek 07TRUNK * 10emc2/tests/ccomp/lathe-comp/expected: these results are right, with the new corrected entry move
EMC: 03cradek 07TRUNK * 10emc2/tests/ccomp/mill-line-arc-entry/expected: these results are right, with the new corrected entry move
EMC: 03cradek 07TRUNK * 10emc2/tests/ccomp/mill-zchanges/expected: these results are right, with the new corrected entry move
EMC: 03cradek 07TRUNK * 10emc2/tests/interp/inside-corners/expected: these results are right, with the new corrected entry move
EMC: 03cradek 07TRUNK * 10emc2/tests/interp/inverse-time-with-comp/expected: these results are right, with the new corrected entry move
EMC: 03cradek 07TRUNK * 10emc2/tests/interp/cam-nisley/expected: these results are right, with the new corrected entry move
cradek: yay, thanks
remember to tell bigjohnt you fixed his bug
he is a great tester
geek night at the pizza+wifi place sounds like fun :-)
yeah, it is
no power here though, so it's limited to 2-3 hours :-)
seb_kuzminsky: I'll make sure and give you 9 hours warning next time so you can come along down
need a caffeine powered generator
haha, "man room"
(I just read back)
oh, pizza, not coffee
coffee was involved...
9 hrs there, 3 hrs until the laptop battery runs out, 9 hrs back, hmm
time to go
their batteries are weak compared to mine! ha!
but now I'm all alone
you're never alone whoaw haw ha ha
* skunkworks tried for creepy and think he conveyed it..
I guess nobody wants to do the dirty work of removing halvcp
I would, but I'm so behind....
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/install/Latency_Test.lyx: adjust image size
BJT-Hardy: dunno if you saw cradek's check in, but he found and fixed that comp bug you mentioned on tuesday
I saw that :)
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/images/HAL_Configuration.png: update image
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/Master_Integrator.lyx: removed 5i20
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (Submakefile docs.xml index.tmpl): remove 5i20 from html
* BJT-Hardy is off to work
BigJohnT is now known as BJT-Work
i'll try to remove halvcp tonight
BJT-Work: you need to get busy finding the next bug for me now
(thanks for testing!)
but you only "fix" every other one :)
really? did I miss one?
no every other one is an undocumented feature :)
I would never have seen this report if jepler hadn't told me
oh, I get it, heh
does g64 treat g0 and g1 the same as far as cutting corners to maintain speed?
hmm.. too bad I missed seb
I started on removing halvcp already
seb doesn't usually get started on anything until late night US time
hope I'll be finished by then ;)
wonder if there's anything else we should be removing
heh, removing isn't a new feature... right? :D
I think creating a halui_pyvcp takes more time than the actual writing of halvcp :/
"halui_pyvcp config" I mean
hmm.. think I found something for the sample pyvcp from stepconf
some LEDs for min/max limits if they are hooked up to parport
homes and limits?
yeah for the actual switches
axis shows if you are on a limit, but not which end
btw, I'm doing the exact oposite atm
building a full VCP GUI ;)
except load program, and preview
not as a stepconf sample, right?
nope, as a halui_pyvcp sample
to replace halui_halvcp
aligning buttons is a pita
[17:17:03] <alex_joni> http://imagebin.org/38247
halvcp has nicer buttons than pyvcp
put the LEDs on the left
I think I'll use a <table> to align them
don't want to make it too wide, or it won't fit AXIS that nicely ;)
you mean the status error leds?
never mind that, it won't ever go in AXIS...
or all of them?
not in AXIS.. but in the sample config
right, do "() Label" instead of "Label ()" and they will look better
alex_joni: use the same <width> for each button to align them
BJT-Work: that's horrible :)
because I have ~ 500 lines of XML now
works nice :D
don't let jepler see that :D
to center the leds look here http://www.linuxcnc.org/docview/devel/html//hal_pyvcp_examples.html#sec:GS2-RPM-Meter
in the manual :)
yeah, I saw the label width
but that's ugly imo
not as ugly as the leds off center imo
seb_kuzminsky: wanted to let you know I started on removing halvcp
I think you need more things open on your desktop alex_joni :)
hi alex_joni, yeah i saw that in the logs this morning
the screenshots look nice btw :-)
anything i can help with? i'll have some time tonight (in 8 or 10 hrs or so)
seb_kuzminsky: don't think so.. I'm just starting removing stuff.. should be straightforward
(if my wifi stops acting up)
EMC: 03alex_joni 07TRUNK * 10emc2/configs/halui_pyvcp/ (README halui.ini pyvcp.hal sim.tbl sim.var vcp.xml): halui_pyvcp config
BJT-Work: feel free to align it how you want
don't make it pink though
ok how about blaze orange
EMC: 03alex_joni 07TRUNK * 10emc2/configs/halui_pyvcp/.cvsignore: silence
* alex_joni likes orange
EMC: 03alex_joni 07TRUNK * 10emc2/src/Makefile: remove halui_halvcp, add halui_pyvcp config folders
alex_joni: in that demo config, can the two guis be in separate windows?
want me to commit it that way?
I think that would be better for the user to understand
ok.. and the user can switch GUIs
yes, also true
EMC: 03alex_joni 07TRUNK * 10emc2/configs/halui_pyvcp/ (halui.ini pyvcp.hal): activate pyvcp in a separate window by default
EMC: 03alex_joni 07TRUNK * 10emc2/src/configure.in: remove mention to halvcp
EMC: 03alex_joni 07TRUNK * 10emc2/src/configure: rebuild configure - mention of halvcp removed
EMC: 03alex_joni 07TRUNK * 10emc2/src/Makefile: don't build halvcp anymore
too bad there are (at least) two ratios of reversing tapping heads. it would be neat to have g84 (or another number) do a cycle for those
can't you specify the ration on the g84 line?
* skunkworks was just going to ask that..
not sure how nonstandard/weird that would be
halvcp is going bye-bye
* cradek waves
there won't be a commit log though
EMC: 03alex_joni 07TRUNK * 10emc2/debian/emc2.files.in: halvcp went bye-bye
EMC: 03alex_joni 07TRUNK * 10emc2/debian/changelog: note that halvcp has been finally replaced by pyvcp
emc2-buildbot mailing list is up
it's quiet when the build succeeds, and sends one email per failed build
[18:56:12] <seb_kuzminsky> http://emc2-buildbot.colorado.edu
seb_kuzminsky: did it mail anything?
whee.. deb builds fine too for emc2_2.3.0~alpha1
guess halvcp is officially gone
it decided to build in the middle of the halvcp remove so the .deb didnt work
seb_kuzminsky: if it's been quiet lately, maybe you can make it email cvs-commit
seb_kuzminsky: sounds like a good catch
yeah, looks good here now that it's caught up with your checkins
I meant to emc-commit
dh_movefiles: debian/tmp/usr/bin/halvcp not found (supposed to put it in emc2)
The requested URL /pipermail/build/ was not found on this server.
it runs the test suite too?
i'd be happy to subscribe the emc-commit list to the emc2-buildbot build failure email list if people want that
cradek: yes, on both rt and sim
failing tests cause it to fail the build and send email
seb_kuzminsky: not warnings though, I hope ;)
[18:59:36] <seb_kuzminsky> http://emc2-buildbot.colorado.edu/buildbot/builders/hardy-x86-trunk-sim/builds/605/steps/runtests/logs/stdio
alex_joni: right, not on warnings
then I'd say go ahead
the ayes have it :-)
there should have been some time last night that a lot of the tests failed
my buildslaves were down last night
moving the vm host to the shiny new machine room we built at work :-D
who runs firstname.lastname@example.org? can the list admin allow posts from email@example.com, so that the buildmaster doesnt have to receive every commit email?
I can add that I think
EMC: 03cradek 07TRUNK * 10emc2/debian/rules.in: normally there aren't any; this is not an error.
or can i have the key to emc-commit?
you can subscribe and set "nomail"
ok i'll just to that
yeah I think that's the easy way
i didnt know that about mailman
if you can't set it, I know I can
but I think you can too - you're supposed to use it when going on vacation for instance.
when turning on your autoresponder :)
I'm working on an autoresponder called "ishouldbeshot"
seb_kuzminsky: managed to do it? or should I?
i got it alex, it's just like cradek said
ok we'll see what happens next time the build breaks
jmk's compile farm and the buildbot should both warn us
there is also [not metoo]
that only prevents you from getting your own email
but you still get the commit email to the buildmaster addy
i've turned off: "Mail delivery", "Receive your own posts to the list", and "Get password reminder email"
shall we test it?
the buildbot email? sure, i could tell it to rebuild an old build that failed
I can commit a borked Makefile
to see the compilefarm complain, and the buildbot
ok, that sounds good
EMC: 03alex_joni 07TRUNK * 10emc2/src/Makefile: intentionally cause a build error. tests infrastructure, will be reverted asap
(hmm.. I think the CF is offline or at least not building atm)
the buildbot checks for new commits on 10-minute boundaries, so we'll have to wait 7
hm, maybe i should teach it to look for mail on emc-commit instead
jmk's polls the cvs server, I don't recall how
it should be in infrastructure
it might have come from not trusting the mailing list
i'm using jmk's scripts to tickle the buildbot too
[19:25:47] <alex_joni> http://cvs.linuxcnc.org/cvs/infrastructure/farm-scripts/check_commit?rev=1.3
those sf mailing lists can be pretty flaky...
but CF hasn't compiled anything since 2009-02-11
at work i run post-commit hooks on my repos that kick our buildbots
we can do that too if needed, since cvs is not at SF anymore
post-commit hooks are used for CIA email
hm, actually now i'm not sure if CU's firewall will allow buildbot-poke packets in... probably not
yeah, but a direct email from the cvs server should be ok
the poll doesn't hurt anything - it's a tiny amount of data
the buildbot does 6 separate builds of trunk, so we should get 6 fail emails in about 2 minutes
sim, rip-rt, and deb, all on both hardy and dapper
there it goes
grid is really nice
(btw.. it responds really slow from here..)
hmm.. now it's ok
i think it gets slow when the cronjob on the buildmaster updates the cvs tree to see if it needs to tell the slaves to build
2 failed already
seb_kuzminsky: might be on my end, or inbetween
I had about 40 seconds response time for the grid page
ugh that's slow
i've noticed the buildmaster occasionally pegs the cpu here...
wonder if we could get rid of those mandb warnings if they are normal
i just got the first build failuer email on the buildbot list
cradek: i'd like that
i looked at the mandb manpage and didnt see anything about telling it to exclude certain globs or anything
so i think that means passing its otuput through grep -v... :-(
kinda gross, but it would green up the reports and make it easier to notice if something really does produce interesting warnings
mandb -q ?
why do some have tests and some not? there's probably a pattern but I don't see it
the rip builds have tests
the deb ones dont
rip-sim and rip-rt
3 build failure emails on the buildbot list so far, none on the emc-commit list, and no bounce messages or anything from the emc-commit list so i think it's just being slow
seb_kuzminsky: does buildbot email when it finishes building all slots?
only when a build fails
hmm.. didn't get an email, and there are 3 fails already
but I did get an email just now that buildbot has been added to the list :/
oops, i forgot to subscribe emc-commit to the buildbot build list
i'll fix that in about an hour, gotta run no
valentine sushi lunch with the wife :-)
ok, I'll revert the Makefile change
EMC: 03alex_joni 07TRUNK * 10emc2/src/Makefile: revert intentional screwup
cradek: passing -q to mandb causes the warnings to go away
that disables all warnings:
*whatis parse for page(sec) failed
*is a dangling symlink
*bad symlink or ROFF '.so' request
*ignoring bogus filename
not sure if losing those would be such a bad thing
ok i added the subscription
redoing just one build with the broken makefile
the fail email went to the buildbot's build list just fine, but the mta got this error from sf.net:
SMTP error from remote mail server after RCPT TO:<firstname.lastname@example.org>: host mx.sourceforge.net [220.127.116.11]: 451-18.104.22.168 is not yet authorized to deliver mail from\n451-<build-bounces@emc2-buildbot.Colorado.EDU> to\n451 <email@example.com>. Please try later.
i think the entity that subscribed to the emc-commit list is "firstname.lastname@example.org", but the From or Reply-to on the outgoing email is the buildbot's build-list's bounce address, which is not an allowed sender to the emc-commit list
i'll unsubscribe the buildbot's build list and subscribe the buildmaster user directly, and tell it to send email to both emc-commit and the buildbot build list on failure
it's hard to test this email thing when SF's mail is down
host mx.sourceforge.net [22.214.171.124]: 421 1b2kzd1.ch3.sourceforge.com: Too many concurrent SMTP connections; please try again later
well, stop trying and maybe it'll clear up :)
fine, i didnt want to send any email anyway
i see now why jmkasunich isn't relying on the commit emails to trigger builds...
well.. commit emails are reliable
emails coming from the emc-commit list aren't ;)
this is sf.net not accepting mail *to* the emc-commit list
"reliable but latent"
yeah.. that sucks
but they quite often do..
commit messages are often late here
"this free service sucks!"
soemtimes one or two out of a set of 10 might be delayed
(from the same commit)
SWPadnos: that's funny in a couple of different ways
dunny "strange" or funny "ha ha"? :)
SWPadnos: I noticed that too
both i guess
followed by a quiet sob
alex_joni, fon't be a fork
or a nerf
fifn't mean to
I figuref that
sure you fo
I thought it was acciffental
whee there it went
(though it technically would be accifdental, since there was no g)
i knew if i just obsessively retried every second eventually it would work
well.. at least they don't queue these
i got it already from sf.net
so it looks good to me now :-)
and I didn't make the Blamelist.. yay?
heh, hmm.. yeah about that
it's always seb's fault
when something breaks
you check it out?
because the cronjob that tells buildmaster to build is run by me
yeah, that's ok
but, my reputation!
guess buildmaster should run on the machine where cvs is?
now when ppl google me they'll see a million broken builds!
and buildmaster should get triggered by the user of the last commit..
I think we don't have that many commits that we won't know who to blame
normally a post-commit hook on the repo triggers the buildbot
and the trigger includes the proper blame list for that commit
I think a "good" way for this is to have the cvs commit hook send email to you (besides the email that goes to CIA)
then you have all the info.. commit happened, who commited it, etc
you mean send email to the buildmaster, and i agree that sounds pretty good
yeah, that's what I meant
what currently generates our emc-commit emails? is it cvstoys?
there's a shellscript that triggers CIA by sending email
and another one for emc-commit@..
guess cradek wrote it
to use emails to trigger the buildbot, i'll have to write a custom parser for cradek's emails
the CIA emails are easiest
they're already in a welldefined format
and you don't care for diff's anyways
check the end of that commit-hook
you'll get an idea how messages usually look like
the $REV for CVS is a timestamp, right?
i dont know what else it could be
I think cradek is sending the last version of the file
revision is commented out, so nothing gets sent
whee cradek's back
seb_kuzminsky: I could easily BCC your address on stuff that gets sent to emc-commit@sf
cradek: I think the cia mail is even easier to parse
assuming esmtp can do that, I've never tried it
CC should work..
i'd like that cradek, to email@example.com
sure, cia is easy
which one do you want?
cia, i think
hold on a sec
echo " <name>EMC CIA script</name>"
echo " <version>1.0</version>"
haven't touched this for a few years now...
how does EMC CIA script 1.0 identify the cvs "revision"? a timestamp?
it doesn't send it if I read it right..
not sure what you mean by revision
me neither :-/
i mean "a string that completely specifies a particular state of the tree"
seb_kuzminsky: well, there is no such thing, except maybe date
a successful commit results in a new tree state, and that's what i'd really like to get my hands on
but what gets in cia's <revision> is ... checking
i think in cvs that means a date+timestamp
cradek: commented out, says alex_joni
oh, right, duh
[22:50:08] <alex_joni> http://cia.vc/clients/cvs/ciabot_cvs.pl
heh, that caught all three of us i think ;-)
that one doesn't send <revision> either
seb_kuzminsky: me twice
cia doesn't say anything about revision in our channel, so it doesn't matter
the "official" cia commit hooks for CVS don't send anything either
unfortunately I don't think I/we have what you want at all
so it says "this list of files were committed" and nothing else, that's ok, we can deal with that
seb_kuzminsky: I guess you only care about the trigger
I thought your goal was to trigger a build
get the email, wait a bit in case there are a few more, cvs up, build
and the author to put on the blamelist
buildbot has a knob to say how long to wait for another commit before starting the build
ok, that address is on the cia Cc:
* seb_kuzminsky eagerly awaits the next commit