Re: Per-request UG(unicode)

php.internals

Sara Golemon

19 years ago
andi@zend.com wrote:
> What soup? Did you check out the patch? > It's actually not bad at all. It's even cleaner than what we had previously > as we just make sure we have two versions and we use the right one... >
I did check it out. Even made changes to try out a few variations on the theme. I also stand by that I don't like the number of variables it throws into the mix. How about a compromise, we finish the code migration effort first, then address per-request semantics either before beta, or in the 6.1 branch. -Sara P.S. It feels a little weird that I'm cautioning you against unexpected side-effects this time (usually it's the other way around between us).

Andrei Zmievski

19 years ago
On Oct 3, 2006, at 9:47 PM, Sara Golemon wrote:
> I did check it out. Even made changes to try out a few variations on > the theme. I also stand by that I don't like the number of variables > it throws into the mix. > > How about a compromise, we finish the code migration effort first, > then address per-request semantics either before beta, or in the 6.1 > branch.
I am afraid that won't work. There are several extensions which need to know the value of UG(unicode) in MINIT hook. This includes ob_gzhandler, I think. Chris Jones from Oracle is waiting on us to finalize this per-request issue before he can start on OCI8 migration. So we need to decide now. -Andrei

Michael Wallner

19 years ago
Andrei Zmievski wrote:
> On Oct 3, 2006, at 9:47 PM, Sara Golemon wrote: > >> How about a compromise, we finish the code migration effort first, >> then address per-request semantics either before beta, or in the 6.1 >> branch. > > I am afraid that won't work. There are several extensions which need to > know the value of UG(unicode) in MINIT hook. This includes ob_gzhandler, > I think.
It's not ob_gzhandler specific, but the output layer initializes several HashTables at MINIT. Regards,
-- Michael

Pierre Joye

19 years ago
Hello, On 10/4/06, Andrei Zmievski <andrei@gravitonic.com> wrote:
> On Oct 3, 2006, at 9:47 PM, Sara Golemon wrote: > > > I did check it out. Even made changes to try out a few variations on > > the theme. I also stand by that I don't like the number of variables > > it throws into the mix. > > > > How about a compromise, we finish the code migration effort first, > > then address per-request semantics either before beta, or in the 6.1 > > branch. > > I am afraid that won't work. There are several extensions which need to > know the value of UG(unicode) in MINIT hook. This includes > ob_gzhandler, I think. Chris Jones from Oracle is waiting on us to > finalize this per-request issue before he can start on OCI8 migration. > So we need to decide now.
There will have many other issues in php6 (related to unicode or not), we simply cannot write in the stone what php6 will be or not at this stage. This is a general comment, not only about this specific problem. As I just asked you on IRC, it would be nice to slow down a little bit, not to slow down the php6 development, but to do not ask us to take such important decisions so early in the game (it is early for me or most of us, even if you started the unicode work a while back). Once 5.2 is out, it will already be better, from a free time point of view. However, having finally tested the patch and tried to understand its impact, I'm also not in favour of a per request unicode semantic (per vhost only would be better, but it is not possible either). --Pierre