[php-maint] Plans for wheezy

[Raphael Geissert]

On Saturday 07 January 2012 11:26:18 Thijs Kinkhorst wrote:
> On Sat, January 7, 2012 01:34, Lior Kaplan wrote:
> > I'm not updated about the wheezy timetables (or general expectation for
> > it), but 5.4.0 is expected soon (RC5 was just released) and would
> > probably have enough time to stabilize (mainly with 5.4.1 or 5.4.2) in
> > order to make it safe for inclusion in a stable release.
> 
> I don't think we have to expect Wheezy freeze around the corner -
> multiarch dpkg isn't uploaded yet and after that's uploaded there will be
> a significant amount of time before the freeze can begin.

http://lists.debian.org/debian-devel-announce/2011/06/msg00003.html says we 
should expect it in June.

From the experience of squeeze's release, I'd rather avoid being on the edge 
when the freeze is around the corner.

> Of course some previewing in experimental would be fine, but you can't
> expect too much testing to come from that. But leave it there for a week
> or two as a first test, then upload to unstable I'd say.

Yes, but for that we need to coordinate the API transition. I have yet to do a 
rebuild of the current extensions against 5.0 (btw, the phpapi- virtual 
package looks funny, perhaps something went wrong when generating it.)

> Having latest upstream branch really helps in security support during
> stable lifetime of PHP, I think.

Yes, that's for sure. We shouldn't even expect a long time support for 5.3, 
based on some of the discussions on php-dev.
The code hasn't changed *that* much, it isn't impossible to backport changes. 
However, in 5.4 there have been a lot of changes to the engine (which is 
hopefully not one of the components often vulnerable.)

> > There's a bigger question of the PHP related ecology (e.g. the big CMS
> > applications) which might be needing a migration or some updates to work
> > smoothly with PHP 5.4.x.

Yeah, 5.4 is going to hit some of them. Some frameworks just need to be 
updated to more recent releases.

> > Of course, having 5.4 in experimental would help
> > this transition. Examples could be SQLite changes (version 3 only) and
> > removing register_globals.
> 
> We shouldn't have anything in the archive that depends on
> register_globals, so this seems like a good thing to me.

I don't think there's going to be any issue with the sqlite change.
I concur with Thijs regarding register_globals. If there's anything left that 
relies on it, it should be removed.

Cheers,
-- 
Raphael Geissert - Debian Developer
www.debian.org - get.debian.net