"Sascha Schumann" <[EMAIL PROTECTED]> schrieb im Newsbeitrag news:<[EMAIL PROTECTED]>... > > you're propably a bit too optimistic, i can hardly imagine that all existing > > extensions all over the world compile out of the box against ZE2. i guess only > > a handful do this. > > Apparently, most of the extensions PHP ships with seem to > work out of the box. I'm under the impression that almost no > changes are required, unless you directly manipulate objects. > That brings up the question -- do we have some documentation > for extension authors which covers necessary changes.
but there are lots that manipulate objects (e.g. all the sql exts have *_fetch_object). they compile fine but those methods won't work and propably crash php. afaik there is no documentation beside the OBJECTS2_HOWTO in cvs. so use the source, luke. > > > the only thing that should be realized is that compiler warnings are still a > > bad thing(tm). i don't see any difference in this compared to changing the > > number of arguments of a function. > > There is an important difference regarding developer > awareness: > > No. of args change: The compiler will always bail out. > Developer knows: Aha, I need to fix something. > > Signedness change: The compiler might issue a warning or not. > Developer knows: Usually nothing. maybe we should change _that_ ? > > There are so many signedness issues in ext/standard alone, > that I would not want to run with such warnings enabled. that's a bad attitude and leads to bad code. i can show you lots of different types of warnings which would result in runtime errors and segfaults, but you surely know them best. but as my patch is already partially reverted and strlen is no longer uint we can stop here too. harald. -- PHP Development Mailing List <http://www.php.net/> To unsubscribe, visit: http://www.php.net/unsub.php
