Julian Foad wrote:
>
> Define the SVN_NEON_0_25 macro always as either 0 or 1 (never undefined), and
> test it consistently with "#if" (never "#ifdef"), so that we can enable the
> "undefined; assuming 0" compiler warnings and get none of them. This is
Here's the alternative:
[[[
Clean up the use of the SVN_NEON_0_25 macro to avoid compiler warnings.
As the SVN_NEON_0_25 macro is either defined or undefined (not always defined
to 0 or 1), test it consistently with "#ifdef" (never "#if"), so that we can
enable the "undefined; assuming 0" compiler warnings and get none of them.
This is the only macro that was being defined and tested inconsistently.
* subversion/libsvn_ra_dav/ra_dav.h
* subversion/libsvn_ra_dav/commit.c
* subversion/libsvn_ra_dav/fetch.c
* subversion/libsvn_ra_dav/session.c
* subversion/libsvn_ra_dav/util.c
Always use "#if[n]def", never "#if", to test SVN_NEON_0_25.
]]]
This has the following advantages: consistent with the way we define and use
our other yes/no macros; doesn't require re-running autoheader when you update
to this patched version; the build system doesn't get those little bits of
duplication that the previous patch added.
Here are the disadvantages: more lines are touched by this patch; Karl says he
prefers the other ("#if") form (but failed to convince me why).
Actually, I now am convinced that this is the better choice, and I'm just going
to do it. It's not worthy of spending any more of our time on it.
r17014.
Phew.
- Julian
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Oct 25 21:33:25 2005