Hello Steph, Tuesday, November 4, 2008, 2:42:05 PM, you wrote:
> Hi Marcus, >>> I'm pretty sure this is a wrong fix - the check for "not shared" replaces >>> the original check for HAVE_HASH_EXT, which is effectively a global >>> equivalent to $PHP_HASH. So it should be something like: >> >>> if test "$PHP_HASH" != "no"; then >>> if test "$PHP_HASH_SHARED" != "yes"; then >>> AC_DEFINE(PHAR_HASH_OK,1,[ ]) >>> else >>> AC_MSG_WARN([Phar: sha256/sha512 signature support disabled if >>> ext/hash >>> is built shared]) >>> fi >>> fi >> >> As far as I can tell that acomplishes the same. Onlz zou > German kb? ;) oops >> changed the >> original logic a lot rather then trzing to keep as much of it as possible. > Nope. The *original* logic said 'is it there?' Greg changed that to 'is it > !shared?' You changed it back to 'is it there?' It needs to be 'is it there > && !shared?' read again. It says if it is shared then issue an error. If it is present then use it. >>> The ext/hash files are already included as appropriate in phar_internal.h >>> and don't/shouldn't need re-including anywhere else. >> >> Doesnät matter. This helps me figuring out what is wrong. > In CVS? >>> Also - it looks like ext/hash needs adding to phar_deps in phar.c (as >>> ZEND_MOD_OPTIONAL). >> >> Zep, that's a good point. >> >>> And config.w32 needs updating to define PHAR_HASH_OK, since there'll be >>> zero >>> hash support under doze otherwise. HAVE_HASH_EXT is at least >>> automatic...! >> >> Since I cannot test on windows I need windows developers like zou to fix >> it. > I haven't been able to test 5.3 in months.. I'll fix/test/merge out of PECL > & 5.2 once it's working under *nix, but currently it doesn't seem to be > working anywhere :) > - Steph Best regards, Marcus -- PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php