Oversight on my part. We specified the symbol name in configure.in, so I slapped the APR_ prefix on there. (the HAVE_ symbols come from autoconf, so we can't do much about it)
How about APR_USING_OLD_EXPAT? I'm not comfy with the APR_HAS_ approach because we aren't advertising the symbol, so we aren't saying APRUTIL has or doesn't have a particular feature. And this is a symbol from autoheader, so #ifdef is the right form. Cheers, -g On Wed, Feb 28, 2001 at 10:39:47AM -0600, William A. Rowe, Jr. wrote: > <soapbox> > > Greg... looking at your new apr_xml.c > > #ifdef APR_HAVE_OLD_EXPAT > #include "xmlparse.h" > #else > #include "expat.h" > #endif > > Can we all agree that APR_HAVE_FOO is _ALWAYS_ defined in terms of a > boolean? > > The reason I broke so much code was grepping for APR_HAVE_ and HAVE_, > expecting > the former to be boolean and the later to be def/undef. (Ok, that and fat > fingers.) > > Not I'm not suggesting that we start populating the public space with HAVE_ > symbols, > and we don't want to conflict with other libraries' HAVE_ symbols. But can > we say > that APR_HAS_ is a def/undef condition, and APR_HAVE_ is a boolean, > expected always > to be declared? > > </soapbox> -- Greg Stein, http://www.lyra.org/
