jepler: I saw you complaining about the unreliability of your dsl modem when it rains...
is that the one hosting cvs?
A script runs every 3 minutes; if it detects that the DSL connection has been dropped, it sends a command to the router to restart it
so it mostly just makes me ping out on IRC
I can handle 3-minute outages
jmkasunich: he gets frantic if it's ever down, so we're pretty safe
cradek (Mr. CVS admin): I want a new cvs module!
yep that's true
a new CVS module? ooh, something exciting
what do you want it called?
we talked a while back about a "misc" or something
for farm scripts, cvs server scripts and config files, etc
no need for misc, we can have as many as we want
jmkasunich: did you see the "fix" I suggested you try to get halscope to build on the farm?
yeah, and I'm gonna try it
btw are the farm results getting put on the linuxcnc.org website again ?
_after_ I make some changes to the farm scripts
the changes will make the farm start within 5 mins of a commit instead of within an hour
I guess I should ask - do you mind if 4 farm slots read a webpage from the server every five mins?
I dislike this date format
jmkasunich: the reason bdi2 succeeds could be that it doesn't have gtk at all, so it doesn't build halscope..
"checking for GTK version... none"
prefer 2006-04-06 I think
I like the makefile change and will do that
06-apr-06 is also unambiguous
I assume thats what you don't like about the other format?
I see it's right on live: Compiling hal/utils/scope_horiz.c -I/usr/include/gtk-2.0 -I/usr/lib/gtk-2.0/include -I/usr/include/atk-1.0 -I/usr/include/pango-1.0 -I/usr/include/freetype2 -I/usr/X11R6/include -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -DGTK_INTERFACE -DHALCMD="/usr/local/bin/halcmd"
when you have international collaboration, I think it's inappropriate
I'd vote 2006-04-06 as well, but I'm not on the board so I assume my opinion counts for dirt <-- humor
jepler: will 1152 hits per day to a file on the server mean anything?
that's probably no worse than a few anxious rss pollers
jmkasunich: that's about how many hits the axis website (also on my dsl) gets a day.
I doubt it will really be noticed
ok, thought so but just wanted to ask
since I'm messing with farm scripts, I can change the date formst
it's a very small file right?
[00:36:22] <jmkasunich> http://cvs.linuxcnc.org/last_commit
1152 hits of a 1k file per day = 1 MB/day = 92 bits/second
I wget and diff it, when I see a change I do a full cvs up
even a 110 baud modem could handle that (barely)
(before I was doing a cvs up every hour)
it's 1xx bytes
it might actually lower traffic overall
depends on the last change of course
oh wrong, it is 210
was 135 earlier
* cradek slaps wget
I left the script (checker only) running overnight, so I was pretty sure I had the latest
anyway, back to the new module
sure, we could have a new module for every little thing, but I'd rather have on module and make trees under it
1) less clutter at the top level
2) can make a new tree without asking cradek
s/on module/one module
misc kind of implies there will be no organization in it
I'm open to better names
or call it i12e
support (except it has more than one meaning)
yeah, and the one the average user assumes is the wrong one
but I think jeff is on the right track
is there a shorter word that means that?
I hope so
utility could be misintrepreted (g-code writing tools could be utilities)
i12e is ugly, but it has the (dubious) advantage that users won't know what the hell its all about
more accurately, they won't make assumptions about what its all about
If it's all the farm scripts, maybe just call it johndeere?
I may end up putting cvs thingies in there too
I also hope cradek will put cvs server scripts in there
CIA mailing stuff, etc
so people can help me (ahem) find the XSS and shell exploit bugs
speaking of that, did everyone see the latest sf site news?
[00:46:52] <jepler> http://sourceforge.net/docman/display_doc.php?docid=2352&group_id=1#1144353894
"Our primary concern right now is for the security and reliability of developer CVS access"
I suggested to cradek that it made me think they know there was or think there might have been some kind of security breach
apparently involving anon access
"Accordingly, we'd like all projects to audit their CVS repository for corruption and raise any issues to us via a Confidential Support Request immediately" -- an earlier message
seems like they have a mess on their hands
they need to come clean about what happened
I suppose we should "audit" our repo there, even tho we don't plan to use it
they don't want to say it, but it sounds like they're scared something got in. Something that should be "Confidential" if you find it.
maybe it's been happening off and on and they didn't do anything.
"On 2006-03-30 the developer CVS server had a hardware issue that required us to take the service offline. "
sure it did
no hardware issue takes a week to fix.
that could simply be corp speak for a crashed disk
once an organization gets above a certain size they forget how to speak clearly
but "wtf happened??" can easily take a week
disk failure -> restore -> someone finally notices the break-in when looking at logs or repository contents -> hush up and take a week restoring the backups
after more time spent with the thesaurus, I like infrastructure; it means exactly what we want
cradek: word geek
why do I lose (NOT LOOSE) at scrabble then?
play against me sometime. I'll be the looser.
LOSER LOSER LOSER
I'm plenty loose after a couple beers.
and I need to be loose before I'd play a game I can't standa
you're plenty loose after one
you don't even have to finish it
wow, weather is happening now
it's been happening over here off and on
it might be picking up again
weather happens constantly
it's happening adamantly right now
just some weather is more happenin' than other weather
how about "buttress"
how bout "infrastructure" ;-)
how 'bout 2006-04-06
does saying 'bout annoy the word geek
nope, but spelling lose with two os does
how about 2006-apr-06?
that's an unusual format isn't it?
06-apr-2006 is more common
unsortable, of course
(I'm assuming date can be persuaded to output three letter months)
I was just referring to the apr vs 04
but sortable is nice
%b locale's abbreviated month name (Jan..Dec)
dammit, I was reading the man page... I'm slow, you have to give me time
I'm slow because I talk too much
I think we should pick 2006-12-01 or 01-Dec-2006 but not some unholy combination
great by me
I sure like it for its sortability
jmkasunich: should all developers be able to edit infrastructure?
I don't see why not
if it's not a security problem, I think yes
its cvs after all
that's definitely true
fsckups can be reverted
ok it's ready then
date -u "+%F %T"
if it's "just" farm scripts, all we committers have a way to run arbitrary code on jmk's machines already.
that's certainly true
only if I update the farm without noticing that you've changed things
no, we just put it in the makefile
jmkasunich: for instance by checking in "sudo rm -rf /" as the first line of configure
and making a configure.in change at the same time so hopefully nobody notices :)
you could reformat it all :-)
"regenerate with different autoconf version"
you mean to the emc makefile?
that will mess up more folks than just me
(but mine will happen faster)
you download "scripts" from cvs and run them automatically
it's true that all cvs users do
you mean configure, or the farm scripts?
the point is, me being able to edit the farm scripts doesn't make any difference to the security of jmk's machine, since I already have a way to run arbitrary commands.
we mean it doesn't matter if people can change the farm scripts because you already have that much risk
ok, but you've also pointed out that I have that much risk, which I dont kile
I'm more likely to write 'if(older than ubuntu) sleep 20 minutes then give a successful result'
I should disable sudo on those boxes
just so jmk stops buggin' me when I break bdi2
oh you have passwordless sudo? that's such a bad configuration choice...
I have what paul set up
* cradek hums to himself
I'm not sure the older ones have sudo at all
I usually find myself using "su -c" or just logging in as root if I have a lot of rooty stuff to do
remember, the goal here was to emulate what joe user would have if he just did an install
I could customize the box left and right, but the validity of the test compiles goes down if I do
joe user has the same security problems...
bdi 2 has no sudo
bdi-tng has it, but the default user isn;t in the sudoers file
it asks for pw, then says "you're not allowed, incident will be reported"
live and 4.20 have passwordless sudo :-(
I understand why it's there, but fixing emc would have been such a better solution
for the farm, is rm /bin/sudo a reasonable solution, or is there something more elegant?
you could just edit sudoers
or apt-get remove sudo
Live isn't debian
or chmod 000 /bin/sudo if you don't want to change anything
rpm --erase sudo
dunno wth flavor that actually is
nor do I care
chmod 000 /bin/sudo then
the right thing would be to edit sudoers
so I can still use it (with a password)
I think morphix is debian-based
HTTP/1.1" 404 295 "-" "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1;)"
my webserver is getting hammered with this crap
"Mambo - the dynamic portal engine and content management system."
"Mambo - the piece of crap security nightmare"?
cradek: I get a pile of those in my logs too
is there a correctly-written php script out there?
cradek: oh probably
10 print "404 not found"
I briefly ran a phpbb and it was the same way
20 goto 10
you guys would probably find the farm scripts amusing (and full of holes)
is the new linuxcnc website being served with apache?
that can do server side includes, right?
depending on the configuration
(we don't have control over the configuration, it's a shared server)
right now, to update the main farm status page, I get index.html, run some sed-fu on it to make changes, and put it back
thats subject to a race condition
hmm, now my rain detector is working
(among other ugliness)
the dreamhost documentation indicates that SSI is enabled for .shtml files. http://wiki.dreamhost.com/index.php/KB_/_Web_Programming_/_CGI%2C_PHP_%26_Databases_/_SSI:_Server-Side_Includes
so the date and "PASSED"/"FAILED" can be included from index.html
well, first things first
you could include the entire <tr>...</tr>
yeah, that would be simpler probably
first things first
gotta import the scripts and such into cvs
yeah .. check them in before you break them
make sure you don't check in keys or passwords!
all that is local to each machine
only the main script and the individual build scripts are in cvs
speaking of which... should the farm be doing normal or RIP builds?
(it doesn't do make install _or_ make setuid, just the make)
I think it doesn't matter much
its doing normal now
the only difference is a few -D... cflags
thats probably better, since most developers test with RIP
I don't see any reason to use --enable-run-in-place
cradek will always "ship" installed, so it's good if it's tested by somebody
so if something happens that works for RIP but breaks for normal, the farm will catch it
cradek: made new debs yet?
oh, that reminds me of another thing to talk about
no, I'm not sure it's the right time anymore
maybe this weekend after we talk about it
if we have a fix in head that should be in the release branch (bugfix) how should we go about it?
um, put it there
is that what you're asking?
if we start merging individual changes over, the ability to do a true merge back into the trunk later gets real messy
(we may not care)
we won't do a merge
the branch is a "stable" branch for this release
development goes on
bugfixes must go in both branches
I don't think there's anything to do but manually apply bugfixes in two places
ok, so for example if there is a fix in head, do a diff of the appropriate HEAD revs to generate a patch, apply it to a working copy of the branch, then test and commit
heh, the cvsweb is nice for making the diff (as long as you know that files are involved)
I want to apply my last rev to scope_files.c to the branch
it changes some shouting "ERROR" messages to much subtler warnings
they aren't really errors
(if you start scope, it tries to restore the previous setup, but if you don't have the pins it was looking at it used to scream ERROR at you)
I can imagine Tillie asking us about that, better to fix before release
arg, cvs gives you some weird crap when you make a new directory, have a look at that commit mail
Modified file infrastructure/cvs-server/-
Modified file infrastructure/cvs-server/New
Modified file infrastructure/cvs-server/directory
well, I was wrong about _which_ special case wasn't handled/tested yet
but right that there was one
jmkasunich: you know the old version was 1.6 and the new version is 1.7, so you can merge those changes into a working copy of the v2_0_branch with this CVS command:
cvs update -j1.6 -j1.7 scope_files.c
-jREV1 -jREV2 merges changes after REV1 including REV2 into the working file
you may get ">>>>"-format conflicts by doing this
I knew that once
I did the merge of pauls lathe fork code (which branched about 6 months before we decided to merge it)
I spent a _lot_ of time reading the cvs book for that
forgot it all since then, that was right after fest last year, probably may
ok, what am I doing wrong:
cvs -z3 -d:ext:firstname.lastname@example.org:/emc co infrastructure
looks right to me
Cannot access /emc/CVSROOT
oh /cvs not /emc
I suspected that part of the line
with sf it was /cvsroot/emc
I know the cvsroot part is gone, but I forget that the emc part is now cvs
tell me when you want to test the "new directory" message
just a minute
ready, I think
john@ke-main-ubuntu:~/emcdev/infrastructure$ cvs add farm-scripts
rlog: warning: `-' is obsolete in `-r-'; use `:' instead
rlog: /cvs/infrastructure/farm-scripts/-,v: No such file or directory
rlog: /cvs/infrastructure/farm-scripts/New,v: No such file or directory
rlog: /cvs/infrastructure/farm-scripts/directory,v: No such file or directory
rlog: warning: `-' is obsolete in `-r-'; use `:' instead
rlog: /cvs/infrastructure/farm-scripts/-,v: No such file or directory
Directory /cvs/infrastructure/farm-scripts added to the repository
well that sure didn't fix it
strange, when I run log<last_commit, it works fine
maybe have to make a new dir in the test module?
I removed the test module
oh, you got rid of it
hmm, I cannot see the problem
well, directory adds aren't that frequent
and a funny mail isn't a real problem
(have you done anything other than cvs admin all week?)
tomorrow night I will
I'd be really sick of it by now
(I felt guilty asking you to add the new module)
it's an interesting challenge since cvs is so crappy
it's just a mkdir...
whats tomorrow night?
friday if I've counted right
I'll do something else
we're going to a baseball game on Sunday
nah, just screw off with friends
home opener is Friday, but couldn't get tickets
are you a baseball fan?
my wife more so than me
we go to 7-8 games a year
maybe you can answer something (asked in all seriousness)
some people have favorite teams that they really love
how do you pick one? seems to me they would all be about the same
there's a lot about sports fannery that I don't get
I'm not a "FAN"
I favor the Cleveland Indians because thats where I live
but do you have any geek-like insight into it?
when I lived in Penna I favored the Pirates and Steelers
ok I can see that
no, my geekery is limited to things electronic and mechanical
human behavior is beyond me
ok, I sometimes have a bit of it in sociology/behavior too
a well rounded geek
ok I'll keep studying that one, sorry for the diversion
want another test case (adding a file)?
not gonna add any more dirs unfortunately
I have about 10 files to add tho
new file works I'm pretty sure
go for it
I'll try this one, if it works I'll do the others all in one bite so I don't spam the list
hmm, must be raining
all my tests show that the new directory handling stuff works right
I get a ping
there it went
took a while tho
msg looks right
if I didn't know better, I'd say you did the cvs add before I saved these changes
which add? new dir or new file?
I did it right before I posted those rlog messages to IRC
and after you said "ready, I think"
so that's when you did cvs add newdir
I figured so, but I can't explain why it still broke
I did the checkout of the infrastructure tree right after you told me it was /cvs not /emc
right, that shouldn't matter, only the cvs add newdir
didnt think it would, just giving you the details
I have only one guess, but my testing says it's fine, but I'm going to "fix" it anyway
nope forget it
ok I know what to do: troubleshoot
ohh, thats interesting
I just committed (added) 10 files
I turned off the email for my testing
actually I changed it to send to me
(it looks like the CIA mail appeared at my console)
the emails look right fwiw
yes you probably got lots of output
you did a set -x or something like that?
I see you added the new dirs to the infrastructure tree
yeah I'll delete them
at SF that meant a major chore to remove them
I guess its not so hard now?
I should have left the test module, but I thought I was done
nah, just rm
beats the heck out of filing a support request and waiting a week
you're supposed to say EUREKA!
AHA I'll get it this time
chris@cvs$ cvs add test6
Enter passphrase for key '/home/chris/.ssh/id_dsa':
Directory /cvs/infrastructure/test6 added to the repository
no rlog stuff, thats good
yeah it's fixed
in your local repository, if you got any test? dirs, rm -r them
I don't (I never did an up)
ok now it's safe then
got another question
I know every committer has a (limited) account on the cvs server
do they have home directories?
yes, because each needs his own .ssh dir
but that's all that' in the home dirs
can they scp files into those dirs?
in the back of my mind I have this idea
that anyone who wants to can checkout the farm scripts and set up a farm box
the checkout is easy
getting the results online isn't
right now I have my boxes set up for passwordless ftp to linuxcnc.org/compile_farm
and dreamhost serves the results
ftp isn't the best way to do things ;-/
you could mail results
I was thinking of scp back to cvs.linuxcnc.org and let it serve the results
mail them to something that will put them online?
thats a thought
I want to get mail working
to mail simple pass/fail msgs to the commit list, and XML to CIA
if I do that much, I can also mail the full log to a webserver
can you describe the problem so I can help?
the mail thing?
I was hoping you'd ask ;-)
I haven't even tried to set up mail since I loaded ubuntu
what's your ip today?
no answer on port 22 today
[02:34:29] <jmkasunich> http://www.wurd.com/settings.php
see the part that says "accessing thru high speed connections"
thats what applies to me
I've been able to get kmain (for example) to receive incoming mail, but not to send
ok they require smtp over ssl
are you 188.8.131.52?
did you try following the instructions (use stunnel)?
the last time I even tried was 6 months plus ago
there's even "to configure stunnel for linux"
[02:37:52] <cradek> http://www.wurd.com/cl_ssl_stunnel.php#Linux
Most recent Linux distributions come with Sendmail compiled with SSL support
well before I was running BDI-4.20
what you want is to configure one of your local machines to use ssl to the smarthost (imailhost.worldnet.att.net)
then let that machine relay for all your other (not as smart) machines
probably makes sense to use the farm router
its on all the time
ubuntu uses postfix which I have nfc about
but google might tell you how to do it
is that the old redhat?
that one is running RH7.2
I think rh72 had a real sendmail setup (rh62 did not)
but there have probably been important patches since then
I guess that doesn't matter, it won't be receiving
you just want to be able to send, right?
don't want it removing my mail from the server ;-)
[02:42:12] <cradek> http://www.cromwell-intl.com/unix/sendmail-ssl.html
hmm, this might be the "other end" of what you want
"answer the X.509v# questions appropriately"
yeah I noticed that too
heh, that's like "the remainder of the problem is left as an excercise for the student"
I'm suffering from stack overflow at the moment
(farm setup, etc, then mail)
when I get some time (maybe tomorrow night) I'll take a stab at it
then ask you for help when I get stuck
I'm not sure that page is what you want anyway
I'll start with the wurd pages
there is this http://www.wurd.com/cl_email_sendmail.php
however that is aimed at dialup att users
who don't have to jump thru the SSL hoop
yeah that's just usual sendmail config, no ssl
(kmail worked fine when I was on dialup att)
I'll still need something like that to start with tho
(I'm a complete sendmail novice, I've heard it was hard to configure and stayed far far away from it)
I've used it forever, so even though it's complex, I don't know anything else
anyway, my agenda for the farm is: get all the existing scripts (and the new every 5 minute thing) into cvs
make new checkouts on the slots
do some mv'ing to put the emc/rcslib/emc2 checkouts under the new farm_scripts checkouts
and get it all working with the existing stuff
is the freebsd box off?
thats why I asked you about your IP
isn't port 22 supposed to hook me to it?
some packets from _and_ to that box were blocked
your initial packet(s) must have gotten thru, or there wouldn;t be any going out to you
have to look at the firewall rules again, maybe I accidently closed the hole
(my default is to reject any packet not explicitly accepted by a rule)
I changed the iptables riles
I was letting in the packet that opens the connection
but not the ones that followed
I think when we first did the install I opened it way up trying to get the ftp to work
couple days ago I tightened it back up
I didn't have the specific rules needed for your ssh, you were going in thru the big hole
can't find my ppmc card
lost in this cesspit I call a basement
in a box on a shelf
no wget on the oldest farm slot
I can't figure out how to make sendmail use smtps to talk to its smarthost... it's possible it can't do that (smtps is an obsolete protocol I guess)
but I do notice that outgoing port 25 connections are not blocked
so you can send most mail (some folks block mails coming from a dynamic IP address)
I really doubt att will accept mail sent to port 25
re: wget - downloaded, but the make won't
struct sockaddr_storage makes it unhappy, investigating now
att won't, but everyone else does
IOW I can use everybody else to spam the world? I thought it would be locked down tighter than that
not really, very few mail hosts will relay today
but to send mail to me at timeguy, your machine can contact timeguy directly
oh, I see
I can send mail directly to a host that I know
well your sendmail will decide where to send for you according to MX records
jmkasunich: bdi2 is based on redhat6, right? there's a wget rpm. http://archive.download.redhat.com/pub/redhat/linux/6.2/en/os/i386/RedHat/RPMS/wget-1.5.3-6.i386.rpm
how is the halui coming?
* SkunkWorks really needs to play with hal and cl more.
I can't remember - did you see this? http://www.electronicsam.com/images/KandT/DSC_0242.JPG
- it is what I would like to get drives for.
phone but looking at image.
Talked with roltek about that tool changer.
We agreed that the old rings on the tool holder is a red-herring as far as HAL
I did to a little bit. he said k&t abandoned random posision tool changing.
I'll write up a spec for the HAL module to handle it properly.
But you can't with this tool changer.
nice - thanks ray.
Selfish motive. I need the module for Mazak.
could with extra hardware - i could think of a few ideas. But I like the rings ;)
one extra column in the table will not be a lot of extra work.
No need for rings. HAL keeps track of pockets and which tool is in each.
That is the way it's done on most carousels without the intermediate arm.
That motor has just a tach on the back or is the tach stacked on the encoder/resolver.
the tach is internal on the end - then a resolver mounted behind that (what you can see)
should be easy to replace with a encoder.
You are going to replace the drives?
amps that is.
MattS has built resolver shaped spindles that take a usdigital encoder.
Uses the same flex coupling and the same mounting as these resolvers.
these are from a diffent cnc machine that we were able to plunder. This is what we want to replace the hydrolic servos on the k&t with
all we got was the servos and the final amp stage (scr's)
A ServoDynamics amp ought to run them easily.
Mounting ball screws may be the bigger job.
we where looking at those also. I sure would rather use servocard/servoamps than step/dir
mounting ball screws?
I spec'd rutex for a project and now wish I hadn't.
I tracked down Vlad, the rutex guy on the phone.
Made him commit to two more years of the 990H. "Oh yea no problem"
and in less than a year, he bails.
Full servo has a real advantage with EMC.
how is the machine running? with the rutex?
set up a m5i20 with little motors.
It does okay. The motors are a bit high impedence for the drive to really power.
Accuracy is good.
accel is a bit slow and the motor is a bit squishy at zero speed.
thats good. what did you mean by munting ball screws? the k&t has some really nice ball screws on all 3 axis's
Oh it was just hydraulic motors to drive em. Sorry.
I converted a hydrapoint that used cylinders.
was thinking of that.
ok - that would be a pain.
Yes it was.
On this one you just need to adapt the motor mounts and couplings.
That shouldn't be to bad.
right - the hardest one is probably going to be z as it doesn't have a servo mounted - but we have not dug into it yet.
Z is under the x table?
no - on top. the run shafts and crap to get the x servo drive up to there 2.5 axis
So X lays on the floor with z stacked on that.
then rotary, then pallet.
And what you have to do is remove the shafts that ran Z and put a motor in there someplace.
right. we have not looked at the blueprints yet to see how it gets up there. who knows - they might be spinning the nut on z
So that's your biggest mechanical challenge.
right - then y as there is not much room. and x "should" be the easiest.
On the electrical/electronic side.
I'd use SD drives or equivalent
The top of the line, usdigital encoders. $220 each
and 2 Mesa m5i20 cards.
Then we'd have to get PeteV or PeteW to write an HAL m5i20-IO module.
why mesa over montec? price?
From there is just wiring and HAL configuratio.
lots more i/o?
Price and the fact that a single spare m5i20 card can handle both jobs.
So $600 gets you 4 axis of servo, up to 4 axes of handwheel, and 96 IO at a minimum.
without the breakout cards.
thank for all your help - I know you do this for a living.
not much living in it these days.
I'm thinking of black box EMC operation these days.
586 slot computer, 2-m5i20 cards, 3 slot backplane.
and all of the control from a palm running linux.
thats pretty cool.
All my shop buddies bitch about the idea of a palm handling user interface.
So some of the tasks like cycle start get buttons.
I use a palm to monitor my car engine ;)
Really! Through the car computer?
(I didn't write it but had some input.)
Will it let you diddle the pid on speed control?
I spose that's hard coded.
The cruise control is a servo system.
With PID calculations.
Some cars come with awful tuning.
no - this is stirictly reading. (can pull and erase check engine codes) - also run solinoid
Lot of "engineering" in there.
I use a separate fuel controller for tuning.
I think I may be at the workshop - maybe fri/sat/sun
cradek talked me into it. ;)
for the swapmeet and demonstartions
Great. I'll see you there.
yeah if you don't want to spend the whole week, that's going to be the fun time I bet
Sunday slows down a lot as folk run off into the night.
Friday and Sat are big days.
I sure would like to be in on the emc classes though (now that I have been reading the schedule)
I don't know if I can pull a week though
You do what you can.
would those classes be taped by chance? ;)
for a nominal fee
There will be handouts. Don't know about taping and how well it would work with a projector.
the hal and ladder logic are "of interest"
we looked it up - just over a 5 hour drive. Not bad at all.
rayh is now known as rayh-away
calibration didn't save one variable in the m5i20 ini file.
Took a long time for me to discover that there were extra whitespace in the .hal file.
i hate issues like that.
how are things?
ok, I guess
I'm trying to install a freebsd
but it's not cooperating at all
I changed the PC, changed the HDD having the same problems :(
have you heard of Coverity?
no, but I know poverty
[18:17:08] <SWPadnos> http://scan.coverity.com/
Hey Steven is back or at least back on line.
create/symlink failed, no inodes free
they have a very good source code scanner
online for a few hours
SWPadnos: does that sound OK for an installer?
only if you have a 100M install partition
pid 110 (cpio), uid 0 inumber 511 on /: out of inodes
this happens after 1.3MB downloaded stuff
I had an 8 GB hdd, now it's 20G
different PC, hdd controller, cable, hdd, PSU, everything
that should be plenty. maybe the partition was made with some weird option that limits the number of inodes
but 1.3M is almost nothing these days
SWPadnos: I tried a single big /, or any other combo to no avail
you reformatted the partitions during install?
yup, installer took care of it
hmmm -then I have no idea
I haven't tried any BSD in years
so - on the coverity thing - I think it may be a good idea to have them do automatic scans of emc
rayh-away is now known as rayh
alex_joni: sounds like the installer is creating a filesystem with a really bogus number of inodes. You should ask cradek if there's some trick; he's a freebsd users.
er, he's just one user
jepler: I already have asked him quite a few times..
and I did try all kinds of setups
well he must be useless then
he has no clue
it's a really weird problem
I can't possibly imagine what's wrong
cradek: not that you have no clue in general, about this you have no clue
and I don't either
Ah cradek's here. Question about ubuntu and "authentication required" message.
I SUSPECTED A FSCKED HDD first
ooops, sorry :)
rayh: are you going to ask your question?
The problem I have is with several packages --gedit for example.
When I open a file there I get this message about authentication required.
It keeps poppin up several times and I click cancel.
I've never seen that, but then I don't use gedit.
what file are you opening?
Then if finally opens the file selection widget.
It does it on any file. The reference is to ftp.lyx.org.
So somehow it things I want to open something there.
I got no answer from the #ubuntu
How are you starting gedit? Are you choosing it from the Application menu?
That or clicking on a text file.
That works fine but when I want to open a second file it does that.
I'm guessing it has something to do with active directory.
but I know nothing.
what's the exact message?
cradek: I can pass some custom stuff to 'newfs -O2' anything that might help?
We'll add the DNS entry for you today. It's covered under your yearly domain
General Manager and Resident Oracle
alex_joni: let's go through it together (later)
alex_joni: you don't typically need to do anything special.
cradek: did you ever use static ip's when installing from ftp?
It is a window that comes upl
rayh: does it look like this? http://emergent.unpy.net/files/sandbox/Screenshot-Authentication%20Required.png
rayh: I got that by creating an incorrect shortcut to ftp.lyx.org on my desktop, starting gedit, opening a file, and clicking the shortcut which appeared in the open dialog.
rayh: it's possible you might get it under different circumstances, such as by gedit remembering the last folder you were in
rayh: I'd start by troubleshooting -- removing a bogus shortcut from your desktop if there is one, otherwise trying to make sure gedit isn't defaulting to a strange location when it opens files, or opening a file that was open in an earlier gedit session
Yep that's real close to it. There is another check box but yes.
alex_joni: not with freebsd6, but yes
Okay. That gives me a place to start.
rayh: do you have any idea how much current would be needed for the high voltage tachometer output?
a few mili volts.
feeds a adc.
sorry. ben a hectic day.
heh - tell me about it
somewhere in there is a voltage divider
I suspect the ADC deosn't directly accept 110V signals
I'm sure that must be true but we are dealing with 60's tech here.
SWPadnos: When coverity first announced their scanner, Python was one of the projects they scanned. I wasn't actually able to see the results of the scans (they were showing them only to "maintainers") but from the discussion on the python-dev list I never got the impression that it had uncovered any "real important" flaws. I also got the impression that they choose the projects to scan, rather than the other way around.
hence the 18.3 volts per thousand.
rayh: Just thinking it through
SWPadnos: do you have some additional experience with coverity that makes you excited about their product/service?
jepler: you emailo email@example.com (I think), and ask them to add your project
jepler: I had heard of the product long ago (it started out as the Stanford Checker)
Yeah, I heard of Stanford Checker BITD.
when they were scanning the linux kernel
their tools can't tell if the scanned software is correct (algorithmically), but they can find a lot of possible errors
yep - they still do
and a lot of the "bugs" are false positives in the kernel
I think the python folks also had a big false-positive count
I talked to a couple of guys from the company, and it seemed like it could be helpful
I'm not sure how many "bad practices" are in emc, vs. outright bugs ;)
argh - the wiki seems to be having trouble
viewing or editing pages?
recently I've gotten errors doing either one, but reloading will work eventually
An error has been encountered in accessing this page.
1. Server: wiki.linuxcnc.org
2. URL path: /cgi-bin/emcinfo.pl/
3. Error notes: couldn\'t spawn child process: /home/groups/e/em/emc/cgi-bin/emcinfo.pl
4. Error type: 500
5. Request method: GET
6. Request query string:
7. Time: 2006-04-07 12:05:16 PDT (1144436716)
ok - 3 or 4 reloads seems to help
that's the same error I get
oh crap - I stll need to register for CNC workshop
I get those 500 errors lately as well.
Time to move the wiki.
it seems that the recent SF server upgrades were downgrades
It does, doesn't it.
hmmm - maybe I should add the dates to the wiki fest/workshop page :)
What's the command I use to see how long it will be before a DNS change gets to me?
Do add the dates.
you can use dig to get lots of info
ok on the dates. are there no plans for the 20/21 dates?
ah dig that's it.
Saturday 20 is the auction. Sunday gets pretty slow by afternoon.
oops - bbiab. time to get the laundry and lunch
SWPadnos is now known as SWP_away
SWP_away is now known as SWPadnos
on that note, it's time for me to go.
catch you guys later. Maybe Tuesday or so
(or afternoon, if thats what it is there)
it's hardly midafternoon here ;)
one last thing - I may want to talk about C++ at fest. just a heads up ;)
heh - that's what they all sya
it has its place I suppose
as long as that place is far from me
you'd be surprised at the places it can be applied
* SWPadnos took some courses on controlling hardware with c++
anyway - fair warning. gotta go.