[PHP4BETA] Re: [PHP-DEV] Re: [PHP4BETA] CVS update: php4

On Wed, 21 Apr 1999, Sascha Schumann wrote: > On Wed, Apr 21, 1999 at 09:34:25PM +0300, Zeev Suraski wrote: > > As it broke PHP 3.0.7-dev under Windows, it broke PHP 4.0 as well... > > Well, if it breaks the compilation, why not simply fix it? I am as we speak, I fixed it in 3.0.7... Fixing it takes time though, so not breaking it in the first place is always a better approach :) > > > php_compat.h is only necessary for flock, if I'm not mistaken. > > and inet_ntoa() Well, until now either we hadn't used inet_ntoa() anywhere (in which case it's only necessary for flock()), or it's supported on all of our platforms. So it should still be in an flock-related file only. > > In that case, please include it only where it's really necessary, and also > > name it appropriately. Right now, including that file from php.h kills > > Win32 compilations. > > Suggestion for a new name? I've renamed it to flock_compat.[ch]] and moved it to ext/standard. > > Also, in general, I want to try and avoid including files in php.h unless > > they're really used by all of PHP. This should make dependencies much > > more reasonable and recompile times much shorter. > > Yeah, but in general, compatible routines don't change that often and if they > change, they affect everything. Since it should be transparent to the rest of > PHP, if it uses a function provided by a system library or our php_compat, it > should not be necessary to include extra files. I don't agree here. Adding files to php.h slows you down whether it's by causing you to recompile files for no need, or by simply slowing down your compiler by requiring it to compile more code. Especially with something that's necessary in exactly one file throughout PHP, there's no justification to include it in php.h. Zeev -- ----------------------------------------------------- Zeev Suraski <zeev@zend.com> For a PGP public key, finger bourbon@netvision.net.il -- PHP 4 Beta