RE: [PEAR-DEV] Expect more!
| From: | Jens Vonderheide | Date: | Tue, 18 Feb 2003 13:45:28 +0000 |
| Subject: | RE: [PEAR-DEV] Expect more! | ||
| References: | 1 | Groups: | php.pear.dev |
| Request: | Send a blank email to pear-dev+get-13447@lists.php.net to get a copy of this message | ||
> Then so be it. In my opinion, it takes more skill to contribute to
> someone else's package or two merge two packages then to just wrote
> something from scratch and add it to PEAR. In the end both the
> developers and the users well benefit more from fewer better packages.
> Ofcourse new packages are welcome as long as they add new functionality
> to PEAR.
I think we have two distinct problems here:
1) The need for a central PHP code repository like CPAN for Perl.
Right now, we have a lot of web sites that have some code, but finding
some existing PHP code for a problem is really a hassle. If I don't find
anything on PEAR and some other selected sites, I usually go and write
it myself. Reinventing the wheel can be much faster than finding the
one shop that has exactly the wheel you need...
That's what I like about CPAN: If I can't find Perl code there for my
problem, I can be quite sure that no such code is available. And if an
INSTALL file tells me to install Algorithms::NPCompleteSolver, I know
exactly where to find it and how to install it.
2) The need for good and reliable classes that developers can use
without much thinking about bugs, backward compatibility and the like.
Because these really are two problems, I suggest two solutions to it:
1) Create a central repository with a common installer (i.e. PEAR) and
let people dump code there freely and without any approval at all.
Any developer who wants to use such code must know that he needs to
check the code thoroughly before.
And yes, let there be 10 implementations of the same solution. If anyone
felt he needed to write a new implementation, even though there already
were some, he would probably have some good reasons for it.
2) Take the best (best being defined in terms of stability, ease of use,
whatever) and mark them such (but them in a special part of the repository,
give five golden stars, you name it). Use the community voting process
suggested in this thread for it.
These classes can be used by developers without much thought and can be
expected to be available on most systems.
Just be .02 Euro,
Jens