Re: [PHP-XML-DEV] [domxml] namespaces / exception / extensionname

Hi Shane

I could not agree more with what you said ;) Therefore I don't have
anything to add to it.  And this mailinglist is maybe the wrong place to
discuss this (and as long as noone voluntarily moves "his" extensions, it
will stall anyway at some point.)

chregu



On Sun, 20 Apr 2003, Shane Caraveo wrote:

> Christian Stocker wrote:
> >><rant>PECL will stop being Siberia only when developers realize the
> >>benefits and choose to support that model of development, and only when
> >>some of them get it through their thick skulls that having the
> >>development in PECL doesn't necessarily mean removal from the PHP source
> >>distribution.</rant>
> >
> >
> > ACK, but until today being in PECL meant being excluded from PHP source
> > distribution.
>
> Sorry to start such a thread here, this has been hashed over and over :(
>   But I'll continue my rant since your so kind to egg me on. ;)
>
> It doesn't mean that at all, and the attempt to make that clear has been
> made but lost in the noise.
>
> <opinion>I for one, think that *experimental* extensions should *not* be
> distributed in the base PHP distribution.  *stable* extensions should.
> There can also be different PHP distribution sources, such as
> php-5.0-stable_extensions, php-5.0-beta_extensions,
> php-5.0-all_extensions.</opinion>
>
> > Even extensions moved from php4/ext to pear/pecl were
> > not distributed anymore.
>
> That is not necessarily by intent, but rather an effect of not handling
> the cvs mapping and tar distribution issues up front.
>
> > And the discussion about providing windows
> > binaries for some PECL-extensions lately didn't put much confidence in me
> > that this will change in the near future (the main point in this
> > discussion that PECL extensions can not be trusted was quite irrational to
> > me..).
>
> The issue is really how to distribute binary builds safely.  Whoever
> made the argument that they cannot be trusted is quite silly (I recall
> something said about someone checking in bad code to pecl, give me a
> break).  It's a location in cvs on php.net.  There is effectively no
> difference in cvs between PECL and php/ext other than location, and that
> using PECL provides better management of releases.  Our current binary
> distribution method is someone checking sources out of cvs, building
> them, and putting them in the phpweb tree.  Without binary distribution
> in PECL, nothing would change.  Someone still checks sources out of cvs,
> builds them, and puts them in the phpweb tree.
>
> > I think, at the end it's some kind of "chicken and egg" problem (yeah,
> > it's easter ;)). Until not one of the main extension of php is moved to
> > PECL, not much will change, I assume. And if one of those main extensions
> > will be moved at all, depends very largely on the maintainer.. Maybe we
> > should move all XML extensions (ext/xslt, ext/domxml, ext/xmlrpc, ext/wddx
> > and ext/xml which got some fixes lately as well) to PECL. This could make
> > some impact on the overall awareness of PECL and the policy of including
> > which extensions in the PHP source distribution.
> >
> > chregu
>
> Which is why I'm kind of bringing this up.  Somebody (with major
> extensions) has to make the leap-of-faith over to PECL to show that
> there is no issue in having code there.  Well, there will be issues at
> first until things like mapping the cvs directory and dealing with
> binary distribution of interim extension releases are done, but that
> wont get dealt with until there is a reason to do it.
>
> And sheep will continue to be scared of the unknown.
>
> Shane
>
>

-- 
christian stocker | bitflux GmbH | schöneggstrasse 5 | ch-8004 zurich
phone +41 1 240 56 70 | mobile +41 76 561 88 60 | fax +41 1 240 56 71
http://www.bitflux.ch  |  [email protected]  | gnupg-keyid 0x5CE1DECB