On Thu, 21 Sep 2006, Erik Huelsmann wrote:
> >I was considering using Kamesh's fix for a permanent solution to the
> >compilation problem I introduced yesterday for pre-1.3.0 APR libraries
> >(which David Glasser fixed, and I spruced up), which boiled down to
> >usage of the APR_VERSION_AT_LEAST(1, 3, 0) macro, which wasn't
> >introduced until about a month ago (in 1.3.0 itself).
> >
> >However, after talking things over on IRC, a more easily maintainable
> >solution was suggested -- SVN_APR_VERSION_AT_LEAST(). This macro will
> >both fix this particular case nicely, and allow us to more easily
> >leverage any future new APIs from APR.
>
> >As this isn't a publicly accessible macro, but does need to be shared
> >across our code base, it was suggested that I put it in
> >private/svn_compat.h. I'm attaching a patch which does that, and
> >makes use of the macro.
>
> Or, as we do with APR_ARRAY_IDX and APR_ARRAY_PUSH, you could just
> define APR_VERSION_AT_LEAST conditionally when it's not defined...
> (saves 4 characters)
>
> I don't have a strong opinion either way, but consistency seems like
> something to strive for.
Okay, I'm committing a change which makes an APR_VERSION_AT_LEAST()
macro available.
How about the svn_hash_clear() function I committed to
svn_hash.h/hash.c a couple days ago? It's basically just a wrapper
around or re-implementation of APR's apr_hash_clear() API (also found
in APR 1.3.0). Should I then do away with svn_hash_clear(), and
conditionally define apr_hash_clear() in svn_compat.h?
- application/pgp-signature attachment: stored
Received on Thu Sep 21 18:45:04 2006