Hello Internals, Johannes, Lukas,
how about moving ext/dbase to PECL, it is absolutely not a widespread ext
and SQLite stuff has prooven to take over local, single file databases pretty
much.
Given the recent discussion on extensions. From my point of viewe there are
two options:
-
Move everything that can be disabled to PECL. This renders the resulting
PHP pretty much useless for allmost everyone and thus forces people to
start using PECL and distributors even more to carefully select. In the
long run this is the way to go anyway and was discussed as such many times
already. Also any ext that moves to PECL should be enabled by default in
PECL of course. A reason to do that move right now would be the approaching
of SVN and the opportunity to go with a clean code layout. -
We might not really be ready for one and continue doing as we've always
done. Select a nice collection of extension that aims to make the majority
of our userbase happy. And suggest defaults this way whether or not they
are enabled by default. The default enabled exts are just a stronger
suggesttion that we think people should be able to rely on.
Best regards,
Marcus
- We might not really be ready for one and continue doing as we've always
done. Select a nice collection of extension that aims to make the majority
of our userbase happy. And suggest defaults this way whether or not they
are enabled by default. The default enabled exts are just a stronger
suggesttion that we think people should be able to rely on.
Yes, having a good feature rich PHP core is important. If an extension
is not enabled by default, implementers of public-ready applications
can't rely on it. Saying that it is "bloat" if more things are enabled
by default isn't really an argument - because the people for whom that
really matters, can very easily use --disable-all.
regards,
Derick
Derick Rethans kirjoitti:
- We might not really be ready for one and continue doing as we've always
done. Select a nice collection of extension that aims to make the majority
of our userbase happy. And suggest defaults this way whether or not they
are enabled by default. The default enabled exts are just a stronger
suggesttion that we think people should be able to rely on.Yes, having a good feature rich PHP core is important. If an extension
is not enabled by default, implementers of public-ready applications
can't rely on it. Saying that it is "bloat" if more things are enabled
by default isn't really an argument - because the people for whom that
really matters, can very easily use --disable-all.
..which actually doesn't disable that much since some people thought making
certain extensions enabled without possibility to disable them was good idea ..
--Jani
- Move everything that can be disabled to PECL. This renders the
resulting
PHP pretty much useless for allmost everyone and thus forces people to
start using PECL and distributors even more to carefully select. In
the
long run this is the way to go anyway and was discussed as such many
times
already. Also any ext that moves to PECL should be enabled by
default in
PECL of course. A reason to do that move right now would be the
approaching
of SVN and the opportunity to go with a clean code layout.
Just to make it clear for me, you are talking about the mid-to-long
term.
- We might not really be ready for one and continue doing as we've
always
done. Select a nice collection of extension that aims to make the
majority
of our userbase happy. And suggest defaults this way whether or not
they
are enabled by default. The default enabled exts are just a stronger
suggesttion that we think people should be able to rely on.
Well this is indeed an entirely different model if we would go with
1). I presume we could offer a "recommended" list maybe. I also assume
that this would shift things towards the linux distros similar to how
things are with the linux kernel. Anyways the first thing is to have
the distribution of PECL packages rock solid. So imho lets wait and
see how pyrus pans out. That being said, PHP 6 would be the best point
in time if we ever want to do this for the next half decade.
regards,
Lukas Kahwe Smith
smith@pooteeweet.org