I don't think I want to add yet another special-case parsing to the
code. I already regret adding the support for CVS $Revision$ keyword,
which seemed like a good idea at the time but now dooms me to have this
one exceptional interface which I can't drop now.
The Subversion revision number is not a version number; it is merely a
monotonically increasing number in the repository history. In other
words, two projects which share the same repo can change at different
rates and their revision numbers have no relation to each other or
really even to their own development history.
Additionally, no algorithm I could come up with to transform the
revision number into a conventional version number (like /1000) would
fit everyone's needs, so I'd hate to pick just one.
If you want to encode the revision number into the release $VERSION (so
you can quickly determine which code a user is running), then add that
to the "real" version number:
# this all needs to be on a single line
our $VERSION = qv('1.2'.qw$Id: filename.ext 123 2008-11-25 08:53:56Z
stro $[2]);
If you are running a Perl < 5.8.x, you need an additional set of
parentheses -- e.g.(qw$blah$)[2] -- because you can't directly reference
the individual list elements (you need to create an anonymous array first).
Using the above code, you can increment the "VERSION" number (the '1.2'
part) at whatever schedule you choose to release the code. But you also
have the revision number as the last term of the $VERSION for quick
reference.
Sorry
John
