Releases, trunk, policy and the wardrobe (Was: Re: RFC: built-in web server in CLI.)

+10000

Only thing for trunk to be done besides stabilization and such is
agreeing on the type hint stuff (and if we can't agree at least agree to
disagree for the moment and get the release out of the door). Currently
we still have the language extension which looks like another syntax but
does nothing in the tree. ('function foo(int $a) { var_dump($a); }
foo("string");' is valid code printing string(6) "string")

johannes

Are people still thinking about this? And how about applying the current/revised patch to trunk thus making it easier to play with and break, but not freeze its features/API yet.

As much as I like this idea — and I really, genuinely,
enthusiastically* do — allow me to open the can of worms and repeat a
question I've asked periodically on IRC and IRL (with varying levels
of directness) over the last few months:

Is there any thought that we might stop throwing things into trunk
soon and start thinking about some sort of release?

5.3.0 came out in June 2009. Even if we start the alpha process now,
there's no chance trunk is going to hit stable until the last few
months of the year, so we're looking at over two years between minor
releases, which is about as long as I'd be comfortable leaving it —
not just for any PR reason, but because there's some genuinely good
stuff in trunk that I think deserves to see the light of day.

I mean, if people want to continue experimenting, that's great, but
should we be factoring this into the thought process, at least?
There's nothing stopping a 5.4 branch being created earlier than has
been traditional (say, alpha 1 or beta 1) and letting development
continue on trunk. In effect, we'd simply be reverting to the three
branch model that's existed for most of the last few years.

Sorry to hijack the thread, but I think the question needs to be asked.

Anviliciously yours,

Adam

• You can insert more adverbs if you like. I am very +1 on this
  feature. I use the equivalent feature provided by Django for
  development all the time, and it's amazingly handy.

-----Original Message-----

From: adam@adamharvey.name [mailto:adam@adamharvey.name] On Behalf

Of Adam Harvey

Sent: Monday, April 18, 2011 8:26 AM

To: internals Mailing List

Is there any thought that we might stop throwing things into trunk soon and

start thinking about some sort of release?

5.3.0 came out in June 2009. Even if we start the alpha process now, there's no

chance trunk is going to hit stable until the last few months of the year, so

we're looking at over two years between minor releases, which is about as long

as I'd be comfortable leaving it — not just for any PR reason, but because

there's some genuinely good stuff in trunk that I think deserves to see the light

of day.

I completely agree.

The current PHP 5.4 source tree (a.k.a. trunk) already has major enhancements. Just look at the NEWS file.

I think it's a mistake we don't start a release cycle. If we try and get too much into this release it just won't happen.

The typehint stuff which has no consensus should be reverted and we should start pushing forward with a release. Things which people wanted to get done and didn’t (e.g. Pierre asked for another 3-4 weeks to do some more Windows work (about 4-5 months ago which I don’t know if ever happened) should not slow down a release and be marked as nice-to-have – if it doesn’t make the train it can wait for the next version; if they do, great.

Btw, just performance and memory use itself is a good enough reason to push a minor release and every day that goes by where we prevent people from getting their hands on it is a day we're warming up the planet faster. This is between 16-20% faster and on average 30% lower memory usage (the latter is the more exciting change thanks to a lot of memory optimization we did; with a release cycle started and people testing/comparing we may be able to squeeze even more out of it). I’d also like to see some of the work around making mbstring a configurable option out sooner rather than later…

My 2 cents. We can work closely with an RM to move things forward and ensure we push out a lot of the really good stuff that was done.

Andi

Hi!

Is there any thought that we might stop throwing things into trunk
soon and start thinking about some sort of release?

Definitely yes. We have some TODO list for that but I think time has
arrived to get it going. The most major things:

1. Scalar typing. I think we need to do a feature branch from it and
   remove it from trunk. If nobody would take it I can do it though it
   probably will take me a couple of week(end)s at least looking at my
   schedule.

2. Traits - I remember there was some unfinished business there?

There's also a number of smaller improvements that we may want in/out
decisions on, and I'm sure people have some TODO items for it, so let's
put them here: https://wiki.php.net/todo/php54

Stanislav Malyshev, Software Architect
SugarCRM: http://www.sugarcrm.com/
(408)454-6900 ext. 227