Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] workspace management question
From: Ralph Castain (rhc_at_[hidden])
Date: 2009-02-19 15:03:56

I think you are all over-engineering what is in truth a very, very
simple procedure I do dozens of times a day...and can best be
"discovered" by simply doing "hg help" or looking at their great
documentation for svn users. ;-)

"svn add" takes as arguments the files that need to be added. If you
do an "svn st", you will see those identified with a '?' at the
beginning of the line. I generally do this whenever I add new files to
the svn/hg repo (as opposed to waiting until I'm about to commit back
to the main OMPI repo) just because it is easier to keep track of
things that way.

Likewise, you will need to do an "svn del" to delete any files that hg
removed. In this case, when you do "svn st", the files will be marked
with a '!' at the beginning of the line.

Unlike hg, svn does its ignore properties on a per-directory basis.
These properties tell svn "ignore these files" so it (a) doesn't show
them when you do a status (i.e., "svn st"), and (b) doesn't track them
or attempt to add them to the repository. We ask that everyone set the
properties when they create a new directory to keep things clean - you
can look at a similar directory to see what should be included.

To edit properties, do "svn pe svn:ignore <directory>".


On Feb 19, 2009, at 12:37 PM, Eugene Loh wrote:

> Terry Dontje wrote:
>> Eugene Loh wrote:
>>> Okay, thanks for all the feedback. New version is at:
>>> UsingMercurial#Developmentcycle
>>> If everyone is happy with that, I'll remove the old version, along
>>> with the diagram.
>> So I like the new text much better than the old, but I think your
>> diagram is still pertinent in the context of the new text. The
>> only issue I have with the diagram is you don't see the steps in
>> the order that are given in the text but I am not sure that is
>> possible to do.
> I think the diagram contributes little to the new text and, in fact,
> confuses things. The diagram really corresponds to the old
> description.
> Jeff Squyres wrote:
>> FWIW, I think your text is generally good enough (some nits
>> below). You might want to say that "this is one way" or somesuch,
>> because there are other variants (e.g., I typically commit
>> multiple times before pushing, etc.). But I doubt it's worth
>> going into that much detail.
> Right. Let experts diverge from the notes as they will.
>> Minor nits:
> Speaking of which, I changed walla to voila. (You said elsewhere
> that you didn't speak French.)
>> - On the "put back to SVN" side, note that you might need to SVN
>> add/ remove as well (i.e., if you added or removed stuff in hg).
> I'm confused. It would seem to me, you should do this after an "hg
> update" in the combo workspace. What arguments should "svn add"
> take? That is, how do you know which files were just added by "hg
> update"?
>> It's also common to need to update various svn:ignore properties.
> What does this mean?
>> - On the "commit to my local hg" side, it's not always a good idea
>> to just run a top-level "hg addremove" -- that may slurp in some
>> kruft files that you have in your workspace. I usually use a
>> finer-grained "hg add file1 file2 ..." approach (because I've been
>> burned by the top- level "hg addremove" approach before).
> Okay. Done.
> _______________________________________________
> devel mailing list
> devel_at_[hidden]