BigJohnT_ is now known as BigJohnT
I'm starting the CVS->git conversion. The CVS is now read-only.
ok git'er done
wait - wait - should we be doing this? ;)
[14:06:14] <SWPadnos> http://www.youtube.com/watch?v=ZSFDm3UYkeE
an interesting google talk on managing open source projects
sounds cool, maybe I'll watch it later
it's about an hour long, so make sure you're comfy
it happens to be by a couple of the subversion developers
[14:10:41] <BigJohnT_> http://www.linuxcnc.org/component/option,com_kunena/Itemid,20/func,view/catid,21/id,457/lang,en/
wow. the opto22 boards just arrived
BigJohnT_: I dunno. as far as I'm concerned, the %-to-% feature should be removed, because all it does is cause occasional problems
but this is about M2
no, he says it's a "program without m02 at the end"
though it's not clear whether he means a bad program or one using %%
as far as I know, M2, M30 or %...% are required
the way I read it, he's saying that he used to get an error message that the M2 was missing, and now (a) he doesn't get that message and (b) he has to type M2 in MDI to get things to work after running such a program
he must mean with %, I just tested and without M2 I get the file ended without...
when I wrap a program with %...% I get a file ended with no percent sign... in 2.4
jepler: I agree the % to % feature is not needed
yeah, 0 and $FILE_SIZE are good markers for the beginning and end of the code :)
I think it ought to be possible to run a file not ending with M2/M30 (which reset a bunch of state). Currently the way to do that is by using %...%
I didn't realize the %...% was different
* BigJohnT_ wanders off to read the manual again
in theory, you should be able to have random text before and after the % signs, and it's supposed to be ignored (IIRC)
I agree that's the idea, but I don't think emc has ever allowed there to be junk before
(but I bet after is ok)
and run several files without resetting a bunch of thinkg
lerman: alex and I set up CIA messages for 'wizards', and I also set it up to send commit messages the emc-commit list.
yep you have other junk after the closing %
does /% work? (ie, selectable program stop point)
I sure hope not
I just tested % ... % and then some random text typed in after the last %
BigJohnT_ is now known as BigJohnT
I bet /% works by pure accident
but, I won't go test it
some things it's better not to know.
I'm thinking of setups like Roland's servo mounts
where you would want to stop early while filling the stations
that's what "Onnn if" is for
sure, or /Oxxx call
open a file with no M2 or starting %, close the error then run it anyway :)
didn't it move from station to station, rather than doing them in order?
so "o if" or "/o call" make more sense
the program was written to do stage 1 then 2 then 3
in trunk it locks up and repeats the error over and over
but of course for four runs, you don't have all the stations filled
SWPadnos: an early end would work when you are starting (filling the pipeline), but not at the end (when it is emptying)
"/" makes less sense since there are 3 stations and you only get two possibilities
I guess it would be possible to make a parameter/var that's a loop counter
if I was doing a high-production version of that, I'd have a switch at each station, use M6x to read the switches, and if to enable/skip each station's code
in trunk if you open a file without an M2 it doesn't disable the run button in AXIS but 2.3.1 does disable the run button
do both report an error?
when you try and open the file yes
this time 2.3.1 ran the file without M2
* BigJohnT wanders back out to work on the Ferguson
steves_logging is now known as steve_stallings
the move from CVS to git is done!
instructions for getting your new checkout are here: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Git
I'm working hard to get wiki pages updated to refer to git instead of cvs
that's a thankless but heroic task
total newbie question - the wiki instructions for git refer to git creating an emc2-src local directory, but the command line for git refers to emc2-dev, is this a typo or correct?
ok, which is your common practice?
I don't know yet :-)
he meant them to both say -dev
thanks, being a newbie I want to be as conventional as possible for a while yet
didn't someone say that you could do a git clone then just copy the created directory to another computer???
sure, you can do that
about to try compiling run-in-place on my laptop, but it is running stock Ubuntu for ease of network access, should I reboot to RTAI in order to compile only (run run EMC)?
not run EMC
no, I don't think that's necessary
and if you don't care about running hardware, you can build with --enable-simulator and then never reboot at all
I'm using eclipse (that might be heresy!) to look through the emc2 code, and I just looked at importing the new git repository with the jgit plugin.
it shows a bunch of branches - what do I want? all of it?
a few of the branches are important, most are varying degrees of old
probably the only branch you will see a lot of new changes on is v2_3_branch
and of course "master"
that being said, I don't know the answer to your question, since I don't know what eclipse plans to do
I don't either. It doesn't show a "master" branch, so maybe that's what it will get anyhow. I guess that's equivalent to cvs "head"?
so what happens if I tell it to get all of the branches, can I check out each branch from my local repo then?
in plain git, yes. You can switch between the branches in just a few seconds, without using the network.
I guess git clone gets all of the branches?
you can always do that -- but I don't know the consequences of the answer you give eclipse.
ok. I'll have to play with it.
steve_stallings: I believe you need to boot to the kernel you will be using to compile emc for that kernel.
EMC: 03jepler 07master: * r61a965cee88c 10/src/ (hal/hal.h hal/halmodule.cc hal/utils/halsh.c rtapi/rtapi.h): use linux errnos instead of custom error numbers
EMC: 03jepler 07master: * r34d2bff6df91 10/src/ (21 files in 9 dirs): Get rid of RTAPI_SUCCESS, use 0 instead
EMC: 03jepler 07master: * r7b9eb0c34c99 10/src/rtapi/ (6 files): get rid of RTAPI_NOMEM
EMC: 03jepler 07master: * r5c6a99bd20ec 10/src/rtapi/ (sim_rtapi.c sim_ulapi.c): get rid of RTAPI_UNSUP
EMC: 03jepler 07master: * r397a9798184d 10/src/rtapi/ (5 files): get rid of RTAPI_BADID
EMC: 03jepler 07master: * rb6f8aae8727b 10/src/rtapi/ (5 files): get rid of RTAPI_INVAL
EMC: 03jepler 07master: * r91d1421fd544 10/src/rtapi/ (rtai_rtapi.c rtai_ulapi.c rtl_rtapi.c rtl_ulapi.c): get rid of RTAPI_LIMIT
EMC: 03jepler 07master: * r896e2272c289 10/src/rtapi/ (rtai_rtapi.c rtai_ulapi.c rtl_rtapi.c rtl_ulapi.c): get rid of RTAPI_BUSY
EMC: 03jepler 07master: * r316790834e38 10/src/rtapi/ (rtai_ulapi.c rtl_ulapi.c): get rid of RTAPI_NOTFND
EMC: 03jepler 07master: * rd628d9c71101 10/src/rtapi/ (5 files): get rid of RTAPI_FAIL
EMC: 03jepler 07master: * r93db610fa0f0 10/src/hal/utils/halcmd_commands.c: get rid of HAL_UNSUP
EMC: 03jepler 07master: * rac62ecdb616f 10/src/hal/utils/ (halcmd_commands.c halrmt.c): get rid of HAL_BADVAR
EMC: 03jepler 07master: * r31dc0db69b8b 10/src/hal/ (5 files in 2 dirs): get rid of HAL_INVAL
EMC: 03jepler 07master: * r9bff015262e6 10/src/hal/ (hal_lib.c utils/halcmd_commands.c): get rid of HAL_NOMEM
EMC: 03jepler 07master: * r497cd263a23f 10/src/hal/hal_lib.c: get rid of HAL_LIMIT
EMC: 03jepler 07master: * r351e5035b083 10/src/hal/ (hal_lib.c utils/halcmd_commands.c utils/halrmt.c): get rid of HAL_PERM
EMC: 03jepler 07master: * r0e6952196129 10/src/hal/utils/ (halcmd_commands.c halrmt.c): get rid of HAL_NOTFND
EMC: 03jepler 07master: * r596666ceb23a 10/src/hal/ (5 files in 2 dirs): get rid of HAL_FAIL
EMC: 03jepler 07master: * r158c1c64ddb5 10/src/ (25 files in 10 dirs): get rid of HAL_SUCCESS
EMC: 03jepler 07master: * rd005a809f46b 10/src/hal/hal.h: kill HAL_xxx error codes
EMC: 03jepler 07master: * r81f1d6c0fd7f 10/src/rtapi/rtapi.h: get rid of RTAPI_xxx error constants
EMC: 03jepler 07master: * re2231f8c5c48 10/docs/man/man3/ (intro.3hal intro.3rtapi rtapi_clock_set_period.3rtapi): got rid of error defines
jepler: looks like we need those .gitignores too
EMC: 03jepler 07master: * r8d29048f8c2d 10/ (186 files in 93 dirs): convert cvs ignores to git ignores
EMC: 03jepler 07master: * raa1e69cced3e 10/ (rtlib/.gitignore tcl/.gitignore): additional files to ignore
I love it: "(186 files in 93 dirs)"
we've traded types of commit floods
instead of getting a flood when one change touches 50 files, we get a flood when one push contains many intermediate commits
yeah, there's one per commit. I think a big flood will be very unusual.
incidentally, I notice it's only ever the 1-line change summary, not the first few lines
that's a feature, no?
yes -- I like it
I think the big floods will go away when the novelty wears off - I certainly would not have done a commit every time I removed one of those error defines - I'd just do them all
hey, my first merge failure :-)
I freely admit that might be a bad use of multiple commits where one would suffice
mozmck - yes, got the warning, now trying --enable-simulator
ok, that goes further, but still does not generate a makefile.inc and same results as best I can tell even when booted to RTAI
fatal: You are in the middle of a conflicted merge.
lots of output from configure that I need to learn to interpret
cradek: are you talking to steve s, or about something else?
I am not getting far enough for the source code to matter 8-)
steve_stallings: you could put your configure output on pastebin.ca for someone to look at
(I'll look, dunno if I'll see anything)
yes, I should, and I am working on that as well as many other basic skills needed to work around Linux
steve_stallings: did you build emc from source on this system before the switch?
steve_stallings: pastebin.ca needs a web browser, I think you already have that skill
EMC: 03cradek 07master: * re21159f45c35 10/src/emc/usr_intf/halui.cc: Make halui show X,Z,W tool length offsets instead of just Z.
EMC: 03cradek 07master: * r6c6cc2a9ebba 10/ (232 files in 98 dirs): Merge branch 'master' of ssh://firstname.lastname@example.org/git/emc2
no, waited until git transition, Matt helped me prep laptop for compiles but we did not get a chance to actually do one
it is raining here to day, so I will work on my skills for a while, now where is that darned USB ethernet adapter so I can get RTAI boot on the internet....
steve_stallings: to be honest, in your shoes I'd focus on getting a sim build to work first
I did, still didn't get make to work properly, give me a few minutes to get organized enough to post from the Linux box
EMC: 03 07master: * r028dd4ead3fc 10/src/ (configure configure.in): fix bold to work properly when not on a tty
cradek: did you notice that the commit message for your "merge branch 'master'...." commit/push starts with "Conflicts:" ? is that something that git generated?
yes it did - I resolved a conflict in that file to allow the merge to work
(I'm a little surprised by the message too)
it automatically noted the conflicting filenames in the change message
it also makes a funky 3-way diff that I'm not entirely confident how to read
++? -+? - ?
ok, half way there, still IRC on old box but managed to post output from configure using Linux http://www.pastebin.ca/1467752
this is not -simulator but that also failed
steve_stallings: "sudo apt-get install xchat-gnome"
steve_stallings: that error looks like one you see on a system where you didn't follow step 2.1.2 http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#On_Ubuntu_with_EMC2_package_already_installed
we're gonna get you entirely on the new box one way or another
oh, OK, but old men have only so many working brain cells, now I gotta learn xchat 8-)
EMC: 03jepler 07v2_3_branch * r604a8036e111 10/ (164 files in 82 dirs): convert cvs ignores to git ignores
EMC: 03jepler 07v2_3_branch * r81ede8d9cc7a 10/ (rtlib/.gitignore tcl/.gitignore): additional files to ignore
EMC: 03 07v2_3_branch * rb0c6632dc1e7 10/src/ (configure configure.in): fix bold to work properly when not on a tty
strange - why did the third one not say jepler?
cradek: I dunno, but both times I committed that fix (on master and v2_3_branch) it did the same thing. I also got 'Use of uninitialized value in concatenation (.) or string at /git/emc2.git/hooks/ciabot.pl line 192.' in both cases.
[17:02:49] <cradek> http://cia.vc/stats/project/EMC/.message/8b6395/xml
($user) = $author =~ /<(.*?)@/;
what in the hell does that mean?
... then you have two problems
give user the part of author between < and @
the shortest part
Author: Jeff Epler <unpythonic.net>
one of my systems must be set up wrong?
yep, that's it
sorry for the trouble :(
strange. should the script check for that kind of error and use something else in that case?
like whatever is before <, or between <>, or something
best to just LART the people who don't properly configure their git
ow ow ow
* cradek tries valiantly to write one line of perl correctly
jepler: that should be fixed to use the full $author if it can't find something smarter
still in newbie mode here, does configure create makefile.inc or should it already exist? http://pastebin.ca/1467752
steve_stallings: yes, if configure completed successfully, it would create makefile.inc
but configure has an error: you don't have development files for "glib"
so it's no use to go on with the "make" step
did you try doing step 2.1.2 from http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#On_Ubuntu_with_EMC2_package_already_installed
ok, I did try to install glib, unsuccessfully, let me look again
maybe it would help to pastebin the output of the 'sudo apt-get build-dep emc2'
I would have sworn that Matt and I installed all the EMC dependencies, but apparently not, back in a few
you can install a different, smaller list of packages (apt-get install emc2-dev) if you only want to build add-on components for emc2
you need more packages to build the whole thing
is libglib-dev or the like on the list of packages it proposed? if so, you're on the right track
ok, sudo apt-get build-dep emc2, seemed to go well, but sudo apt-get install build-dep emc2 failed with E: Couldn't find package build-dep
that's a typo -- was that on the wiki?
the first command should have done all that is needed
sudo apt-get build-dep emc2
sudo apt-get install build-essential
the wiki shows two lines to run in section 2.1.2
apt-get build-dep <some-package) means "get everything that is needed to build <some-package>"
apt-get install <some-package> means to get that package itself
build-dep is not a package
maybe your hands were thinking too independently when you typed in the second line?
build-essential is - its a meta-package that gets you the compiler and other basic stuff
ah, my mistake, cloned line and did not edit!!!
second command from wiki now says I am up to date and good to go 8-)
then try configure again
and don't bother going on to "make" if it seems to end with some kind of error
got makefile.inc, make now running
really intended to get this stuff done at the fest, but the best laid plans of mice and old men...
seems to have run, got a few warnings, see http://pastebin.ca/1467799
you talking to me, and is that good or bad?
was that build for simulator or realtime?
that looks approximately right
you have my permission to do the next step, sudo make setuid
no hardware here, but the last working ini that I created at the fest seems to work with run-in-place
THANKS to all!!!!!
hm, is this important? if so, we'll have to figure out what to do about it: ./include/rtapi_common.h:static char rev = "$Revision$";/* magically updated by CVS */
git isn't going to make a $Revision$
there is probably another way to solve the problem that addresses
which is making sure that comps are all using the same version of critical data structures
EMC: 03jepler 07master * r39c6a375a13a 10/ (11 files in 7 dirs): get rid of a lot of mentions of cvs
jepler: you seem to have gotten rid of a lot more than just mentions of CVS
like, the list of options that can be passed to configure
that list was out of date
the suggestion to use --help is better
I left that bit, didn't I?
likewise, the suggestion to *not* use --enable-run-in-place is probably a mistake for most users, so I killed it
it wasn't really a related change, it's true
I missed the --help part
EMC: 03jepler 07master * r777e6974a2cb 10/docs/src/ (8 files in 5 dirs): get rid of more references to cvs
steve_stallings is now known as steves_logging
EMC: 03jepler 07v2_3_branch * r3951a192f2f5 10/ (10 files in 7 dirs): get rid of a lot of mentions of cvs
EMC: 03jepler 07v2_3_branch * r2ac36f78a9d3 10/docs/src/ (8 files in 5 dirs): get rid of more references to cvs
EMC: 03jepler 07v2_3_branch * r7aae7b357b11 10/tests/overrun/ (checkresult test.hal test.sh xfail): make a proper test for overruns, instead of an "expected to fail" test
is the sandbox still working?
I get "fatal: The remote end hung up enexpectedly" after a pull
or is that because I did a git clone git://... instead of a git clone ssh:/...
no, the sandbox is gone
those error messages suck
yes they do
BigJohnT: unlike the change from experimental -> sandbox, I don't think there's a way to switch short of doing a full download
I just had the slap on the forhead moment with git branches
I checked out 2.3 then opened a file and edited it then commited it
"git branch" will show you the branch you're on
then I switched to master and opened the same file up and it was as I edited it while in master
I couldn't figure out for a while where the 2.3 branch was :)
it's magic I think
so editing for one branch changed it in others?
oh, I see what you're saying...
there is a way to update the branch from the master but it excapes me at the moment
if you want one change, then you can git cherry-pick
in gitk, you can right click on the change you want, and "cherry pick" it. That will make the same change on your current branch.
make the change on master, go to the branch, then cherry pick it
then, you can "git push --all" to share the whole works at once
you can get to gitk from git gui that is cool
ok, cool that works easy
so now I think it is better to use lyx 1.53 :0
* BigJohnT runs and hides
I can be in master, open a file, checkout branch, open the same file up and I have the master and the branch open... it's magic :)
I wouldn't bet that "cherry pick" will work right if one version of the document is written in new lyx and one in old lyx
they move all the words around, break lines differently, use different codes
it won't work automatically
jepler: once you open both up in 1.53 and save them there should not be a problem right?
BigJohnT: we still can't do 1.53 for v2_3_branch, I didn't work that much magic
jepler: is that because of Ubuntu 6.06?
jepler: if it is not too much trouble can you revert it so I can do master docs with lyx 1.53 and 2.3 docs with lyx 1.3.7
BigJohnT: if you like, yes. hold on...
EMC: 03jepler 07master * r2570159b855e 10/src/ (configure configure.in): switch to lyx 1.5.x for documentation
BigJohnT: I think that does it.. holler if you're not satisfied.
jepler: thanks for doing the wiki work. looks like a huge list.
cradek: I hope I didn't mangle the pages too badly