Re: cvs: php4 / acinclude.m4

On Wed, 20 Sep 2000, Jani Taskinen wrote: > On Wed, 20 Sep 2000, Sascha Schumann wrote: > > >On Tue, 19 Sep 2000, Jani Taskinen wrote: > >> sniper Tue Sep 19 16:52:24 2000 EDT > >> > >> AC_DEFUN(PHP_X_ADD_LIBRARY,[ > >> - ifelse($2,,$3="-l$1 [$]$3", $3="[$]$3 -l$1") > >> + if test -n "$1"; then > >> + ifelse($2,,$3="-l$1 [$]$3", $3="[$]$3 -l$1") > >> + fi > >> ]) > > > > Why don't you fix the caller? The build system is completely > > under our control, so I'd prefer to stop passing bad values > > instead of adding a check to every invocation of heavily used > > macros. > > Well, I took the easy way. But I still disagree with you. > If some macro/function is faulty then why should every instance using it > be fixed? If you can get away with ONE check why adding several ? This is nothing you would have to check for heavily in *every* config.m4. The library list is usually static and only dynamic in a very few cases (oci8, db?). And especially the bug you are trying to fix seems to be caused by only one config.m4. > Hmm..it might be a good idea to have something about build system > in CODING_STANDARDS ? As it looks like every config.m4 there is very > differently build up. So it is quite hard to go and modify them. > And a doc about the build system internals would be nice too.. Sure, if you want to write it, go ahead. > As it is always stated, you should look for examples in ext/mysql > when coding your own extensions..but the config.m4 there isn't so 'nice' > or compact to be example. It's more like do-not-do-it-like-this example. > > So which ext/*/config.m4 is the best one for example? The one which is used by ext_skel. - Sascha