SWPadnos_ is now known as SWPadnos
EMC: 03cradek 07TRUNK * 10emc2/src/hal/components/mux4.comp: typo
EMC: 03cradek 07TRUNK * 10emc2/src/hal/components/mux8.comp: mux8 component contributed by s. stevenson
it seems when I created the "new-teleop" branch back in eary march I forgot to tag the branch point
makes it a bit inconvenient to see what I actually changed in the branch
I think I have it pinned down by date though
I never tag branch points
I don't know what cvs's problem is that it can't give you a diff for the branch
maybe it can and my CVS skills are just rusty
cvs up -jbranch gives you just those changes ...
so you CAN get it by doing that on trunk and then doing a basic diff
so I bet there's a way
what I'm actually trying to do is to move my work on the new-teleop branch to alex's joints_axes branch
oh, that's easy
it doesn't make sense to have those two running in parallel
get his branch (cvs up -rhis) and then merge yours (cvs up -jyours)
the last command is executed while inside his, right?
then fix the conflicts :-)
cvs up -rhis sets sticky "his" branch tags on everything
I already have a checkout of his branch
jmkasunich@mahan:~/emcdev/emc2jointsaxes$ cvs up -jnew-teleop
cvs [update aborted]: no such tag new-teleop
I know the branch exists and that is its name: http://cvs.linuxcnc.org/cvs/emc2/src/emc/motion/command.c?graph=1
cvs log some-file, note how the tag is spelled
telop not teleop
staring me in the face
3 files with conflicts ;-(
I only had changes in 5
they'll probably be minor
heh, in at least 1 file its just the title block stuff
that's so irritating
I always just delete those stupid things when they bite me
Last Edited By: $Look at the log, moron$
now that it's too late, I think cvs up -kk -jyours would have prevented that mess
2 of the 3 were title blocks, one was real
the last change stuff is useless
I wonder though, it might be good to have the revision number in there - for the benefit of people who are using a tarball instead of a checkout
if we really care about that, we should put it in ident strings
No manual entry for ident
that way you can get the information from a binary too
apt-get install rcs
I'll pass on rcs
the whole thing gets messy anyway - cvs and rcs both make versions on a per-file basis
[01:43:06] <cradek> http://netbsd.gw.com/cgi-bin/man-cgi?ident+1+NetBSD-current
when you link a bunch of source files into one binary, which version would you use>
one version for each file
I get it - each file would do something like "static (const char) *ver = "$blah""
I still think it's kind of silly, but it would at least be maximally useful if we want to deal with the silliness
well, we have a mix already, some files have it and some don
I'm certainly not volunteering to go in and add it to all
(or remove it from all)
for me it's only a minor annoyance and if I remember my -kk, it's no annoyance, only silly
so I guess I don't really care
considering how much trouble I've had getting myself to work on emc lately, (as opposed to simply using it), I'm gonna work on things that matter a bit more
the new jogging will be great
my concave comp stuff is 60% done. the rest is no fun. as you might predict, that means it has come to a standstill
EMC: 03jmkasunich 07joints_axes * 10emc2/src/Makefile: merge the 'new-telop' branch into 'joints_axes', two branches for such similar work is dumb
EMC: 03jmkasunich 07joints_axes * 10emc2/src/emc/motion/ (command.c control.c homing.c motion.h): merge the 'new-telop' branch into 'joints_axes', two branches for such similar work is dumb
SWPadnos_ is now known as SWPadnos
cradek: yeah, but cvs up -jbranch doesn't give you a diff against the branchpoint
it gives you a diff against the current version
that's why I tag the branchpoint (in my mind I think it's easier to merge that diff to something else..)
alex_jon1 is now known as alex_joni
alex_joni: I'm pretty sure you're incorrect about what cvs up -jbranch does
it determines the common ancestor, which is the branch point
I only wish you could ask cvs diff to calculate the common ancestor for you too
cradek: I have no idea what -jbranch does, but if it does what you say, then it's very nice
(and lots easier)
were you talking about diff or update/merge?
as far as I can tell, cvs does the right thing for update/merge but I can't figure out how to get a branch diff
I think I can write a program to create a mergepoint tag after the fact
if that would be helpful
it involves looking at the cvs log, finding the branch tag, and chopping off the last two parts (e.g., 188.8.131.52 becomes 1.8)
it was not needed in this case. what jmk really wanted was up -j, but he didn't know it was as easy as it is
he thought he had to produce a diff himself, and apply that
I wrote it anyway, though I am too timid to test it. http://emergent.unpy.net/index.cgi-files/sandbox/make-branchpoint-tag.py
since actually running it would invoke cvs many many times, probably nobody else should actually run it either
your script adds a tag at every branch?
it helps in the case where you created a branch tag but not a branchpoint tag; it's difficult to get a diff from the start of a branch till now unless you created a branchpoint tag
is it time for a revision control system flamewar yet? :-p
no. anyway, I don't think you'll find anybody who thinks that CVS is the best revision control system of all. It's merely the one we use.
have you seen launchpad? http://launchpad.net
it's like sf, but how you'd do it in 2006 instead of 1996...
might be worth at least talking about upgrading the project infrastructure
which parts of launchpad's functionality do you think would be particularly beneficial for us?
import of cvs trees, bzr integration (mostly useful to us for the branches), bugtracking, maybe the forums
but mostly import of cvs & then use of bzr
considering that none (or one) of the developers uses bzr, I'm not sure it's a great advantage :)
unless it really does a lot that cvs can't do (and that git and subversion also can't do)
if we decided to change revision control methods, I would still prefer to self-host the main repository.
bzr does branches and merges well, like git and hg but unlike cvs and svn
jepler: you can still do that, launchpad will happily just have a url to your repo
in fact that's the normal mode, with optional mirroring at their site
there is no "main repo", there are only branches. branches can be anywhere (including at launchpad)
someone would host a branch we all agree to treat as special and call "TRUNK" or something
anyone wants to branch off that they just branch, storing their branch on their computer and/or at launchpad
if they want to share they can register their branch with the emc launchpad project, and others can see it, diff against it, merge from it, etc
I understand that there's not a special "main tree" to the bzr software, but there is by virtue of the project organization
the "main tree" of the emc2 project would be treated much like the TRUNK branch now, with strict access control and stuff
I think branches and merges is one of the things cvs does just fine. I use them all the time.
this conversation started when I went to look at jepler's script for branchpoint tagging... it struck me as a workaround for a deficiency in cvs
If the brancher wants that tag, he could create it when making a branch. Maybe it is only a training issue.
I might do that next time, not sure.
EMC: 03tissf 07TRUNK * 10emc2/docs/src/hal/drivers_fr.lyx: French translation update
I spoke with Roland of CNC-WORKSHOP fame last night. He has had some family and work related issues but promises to have the web page up and running by the middle of next week. They will begin accepting apps then.
ok. thanks. I was wondering if things were all right
He is most anxious to demonstrate the Mazak and has collected a bunch of tool holders. I believe that he plans to do the gcode and such for the milling and tapping of the motor mounts.
oh - the guy who asked for someone to do a presentation at this conference lives in Peoria. he may pop out to the workshop
We will have space, much like last year for developer work as well as class and display/demo/work areas.
That would be great.
well, gotta go prepare for the presentation. see you later
There you go.
you going to the workshop this year?
Do we have any further recommendations about space and equipment that he provides.
not at this instant, but maybe later
this may be of interest to you: http://www.micromint.com/products/electrum200.html
The developer area was okay last year?
they use linux+RTAI by defailt
I think so. I can't think of what might have been better, other than a separate circuit for Aram's coffee maker :)
Ah okay. I'll bring a few 480->120 transformers.
just get covers for them this time ;)
just ask, and ray will bring it
ray can we have some 100hp servo's to play with?
My car will be sitting as low as that honda you took the atlas home in.:)
I've got 25 hp.
ok. really running now
JMK would be better for the big stuff.
I had it temporary. To prove a point at the shop.
Old style open frame motor. Not really good for serious servo except 30-120 percent of nameplate.
Do you think there would be an advantage to setting up a lab with 8-10 PCs networked to a Sherline or two?
8 sounds like a lot, maybe 3 or 4
but if there's space, i dont see why not
Seems like it would allow us to run a whole bunch of workshops for small groups.
If we had one subnet server we could vnc into all of them and project any display.
rayh: did you see http://www.youtube.com/watch?v=mxxdq6y8z8M
I'm glad to hear you're coming back this year.
What no auto tool change?
Wow. That is some impressive motion.
the drilling at the end is canned cycles!
Does it properly move around the tool tip and allow tool length and diameter offsets?
length yes, diameter no
Certainly got the vector motion correct for drilling and such.
yes that's G17.1 G81 W-...
W is always along the tool vector
That is some really great stuff. Is it linked in the wiki and org.
don't think so. he just put it up this morning.
Wow. The features just keep coming.
That same setup would work well for the cable hexapod.
I'd bet NIST would loan it for the fest.
that looks like something out of a sci-fi movie
what's up ray?
rayh: gotta run for 20 minutes.. talk to you later