clkao, who does a fantastic job of maintaining our SWIG/perl bindings,
has the beginnings of a new version control system written. It's called
"SubversionKeeper" (svk), and it's modeled somewhat after Arch or
Bitkeeper. It's a decentralized network of repositories, and users
"push" changesets around at each other using the RA layer directly. The
new system doesn't use libsvn_wc at all -- clkao has written his own
lightweight working copy system. And to top it all off, he's even doing
some degree of "smart merging", i.e. using properties to track merges.
Really impressive stuff.
Anyway, clkao wants to do some more maintenance on our SWIG/perl
bindings, but he'd like to "push" his changes at our repository using
svk. In other words, he's not going to be committing from a normal svn
working copy. He's just going to push a changeset from his private
repository to our public one. I'm not particularly worried about that,
since he's using the same RA commit interface that 'svn commit' uses.
But here's the rub: when svk commits, it sets a custom property that
records some merge tracking data. It looks like this:
svm:source = ae6c956b-9dc6-0310-97b2-e73af4192982:/svn/local:6908
Now, as far as Subversion is concerned, this is a "user" property. It
may as well be "color = green". It's not in the svn: namespace, so it's
just random user data.
The question is, are people going to freak out if they see a property
like this appear in our tree? Most people seem to have an initial gut
response of "ewww, get that crap out of our tree".
But I'd like to argue the other side: I don't think there's anything
wrong with this. If that same piece of metadata appeared inside a
Makefile specific to the SWIG/perl bindings, nobody would care at all.
It's just some data that the maintainer is using to help maintain an
area of code, right? My feeling is, text or property, it doesn't
matter. It's just part of the SWIG/perl code.
To unsubscribe, e-mail: firstname.lastname@example.org
For additional commands, e-mail: email@example.com
Received on Wed Nov 5 17:44:35 2003