Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:51150
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain metux.de from 82.165.128.25 cause and error)
Date: Fri, 31 Dec 2010 12:49:27 +0100
To: internals@lists.php.net
Message-ID: <20101231114927.GC18520@nibiru.local>
Reply-To: weigelt@metux.de
References: <AANLkTi=ox5uUvEpo8ZuegpTPJLiS353Oqtud1WFGOPUT@mail.gmail.com> <1290504719.2294.251.camel@guybrush> <AANLkTi=Lqki0nY2wgKRafTBGinPZZ+VLqkHagXzyJiMq@mail.gmail.com> <AANLkTimrd_5do=3PLB4w1jBNzzO0LBY4Jvqftjqeda7X@mail.gmail.com> <20101231035937.GA18520@nibiru.local> <4D1D695C.9040400@sugarcrm.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <4D1D695C.9040400@sugarcrm.com>
User-Agent: Mutt/1.4.1i
Subject: Re: [PHP-DEV] [RFC] Release Process
From: weigelt@metux.de (Enrico Weigelt)

* Stas Malyshev <smalyshev@sugarcrm.com> wrote:
> Hi!
> 
> >Just had such a problem myself a few weeks ago: a minor update
> >(IIRC was from 5.3.2 to 5.3.3) broke virtually all of my web applications
> 
> Out of curiosity - what exactly broke it? What change was it?

IIRC many deprecation warnings, which totally broke the output.
I guess with the next release I can expect certain warnings to
become true and certain functionality disappearing ;-o

> >I got the strange feeling that php-devs don't care very much of long
> >term stability ;-p
> 
> If enough people would actually test pre-releases, the probability of 
> this happening would be lower.

I, personally, have no time to test prereleases of dozens of 
packages (php is just one of dozens I'm using, not in the list
of those I'm actively developing on). I use the packages of my
distro (Gentoo), which has an own release process where new
packages first masked as unstable and only get stabelized after
some testing period - even this didn't catch this. Even large
distros cannot test everything.

IMHO there's a fundamental flaw in the development process:
existing semantics in a stable line should _never_ change.

Things like certain superglobals or calltime-reference passing
should neither disappear or suddenly produce warnings within
the stable line. The whole idea of depcreation warnings which
explicitly have to be switched off is insane from the operator's
perspective - it means a semantic change which requires each single
php application to be tested in realworld scenarios before rolling
out a new php version to production servers, leading to imensive
costs simply driving people away from php completely.

I really can understand the idea of these warnings, as well as
removing things like superglobals, from development perspective.
But they better belong into an additional instrumentation system
which does not affect the behaviour of the running applications
(eg. send these warnings to syslog instead of stdout, unless
_explicitly_ asked to do otherwise).


cu
-- 
----------------------------------------------------------------------
 Enrico Weigelt, metux IT service -- http://www.metux.de/

 phone:  +49 36207 519931  email: weigelt@metux.de
 mobile: +49 151 27565287  icq:   210169427         skype: nekrad666
----------------------------------------------------------------------
 Embedded-Linux / Portierung / Opensource-QM / Verteilte Systeme
----------------------------------------------------------------------