Re: Re: cvs: php4 / TODO-4.1.txt

At 01:22 18/4/2001, Stig Sæther Bakken wrote:

I'm replying to another message you sent where you say you fail to see the advantages. Well, the advantage is that when you free the PHP core from the release cycle of all the extensions, doing PHP releases will become less complex, the QA process is broken down into smaller, more manageable tasks.

As I said, I don't think this is necessarily a good thing with the popular extensions. It's a great thing with the 'moving targets' or with the not-too-popular extensions, but keeping the release cycle of modules like MySQL, XML, regex, etc. in sync with the PHP release is a good thing in my opinion. Most people will only be bothered to upgrade their PHP installation when there's a whole new release that's out.

Also, it makes the development cycles of the core and extensions more free: if Thies wants to do some heavy surgery on the oci8 extension adding support for Oracle 15, he doesn't have to wait until the "now-in-RC" PHP is released, he can just go ahead and code knowing that the PHP release will ship with the latest version of the OCI extension that he labelled as stable.

Well, that's true, but I think that PEAR isn't really the solution to the fact that CVS has sucky branches (which would have been the right solution for this problem :)

We won't end up with no extensions but ext/standard in PHP, but I think we should be offensive about taking them out unless they are very tightly integrated with PHP, and unless the maintainers object. The only ones I can think of that are this integrated must be "standard", "bcmath" and "session".

At least I get to keep MySQL and PostgreSQL in :) Zeev