Hi,
ok, just came home from some family business and on my way to bed, but
some general comments from my side, I will answer to individual things
tomorrow morning my time.
5.3 was branched off September 2007 and we aimed for releasing it
multiple times since then, just some random comment I recently saw on
twitter:
Quoting @smalyshev:
Old mails: me in Nov 2007: "5.3 is some months ahead, chances of
having next 5.2 are quite low". Shows how good I am at
predictions...9:23 PM Jun 23rd
Often at these times PHP was in quite good shape but we decided to add
this tiny feature here and that tiny patch there and then came some
bigger things like closures and changes to namespaces, and, .. and we
delayed and delayed. Meanwhile it is 1.5 years later and looking at
UPGRADING and NEWS 5.3 became one of the biggest releases ever.
And at some point we have to release and get it out - the list of, moe
or less, planned features is already quite long. This lead Lukas and me
to the decision to become stricter and stricter about stuff as we agreed
that it's better to have some known things not perfect than too many
things introducing bad side-effects - and better than waiting for the
100% perfect thing as that will never be reached. (there is no bug free
software, well maybe TeX, but I'm not certain there either)
We were transparent on that on internals and I spent quite some time
looking at every commit over the last weeks and discussing every change
and got mostly supportive responses - some disagreed with some decisions
here and there but I had the feeling that the goal was shared.
Now we tagged and packaged a 5.3.0 quite some time ahead of the
targeted release time, which again was communicated via internals. One
main driver there was that by this we have some time to test the final
selection and do different kinds of QA. Of course doing tests and QA
means that issues may come up which means we might have to react*) -
while we should make sure that only happens in bad cases as every change
adds new risk (either new bugs or some more long taking bug fixing which
is annoying for everybody who has new features ready, so we might accept
little features again, followed by stabilizing of them, which again
means we release a few weeks later and have a bigger and more complex
release which needs longer testing which ...)
As said in the introduction: I'll answer to individual items tomorrow,
now I just wanted to make sure nobody thinks I packaged something and
left ;-)
johannes
--
PHP Internals - PHP Runtime Development Mailing List
To unsubscribe, visit: http://www.php.net/unsub.php
