Re: Re: [PEPr] -1 for Text::Text_Huffman

Zitat von "Tomas V.V.Cox" <cox@idecnet.com>:

Jan Schneider wrote:

Zitat von "Tomas V.V.Cox" <cox@idecnet.com>:

IMHO we should encourage PHP5 only packages.

Bad idea IMHO, at least at the current point of time. PHP 5 was released today, but it will be weeks or months until you could call it stable for production evironments seriously. And it will take months, and probably a *lot* of months until a majority of ISPs will have switched to PHP 5. Until then no enterprise user or other people depending on a production level environment will be able to use this package.

See my point Jan. LAMP is a very succesful platform for web development, yes? That involves Linux and it involves distributions war. Part of that war is to pack the very latest version of key packages, this includes PHP. I bet you, in less than one month, one major Linux distrib will ship PHP5 and a lot of developers will just start using it. ISPs takes a little long to upgrade, but not so much, for not saying the amount of people who doesn't depend on them.

My personal experience is a different one. I agree, most distributors will include PHP 5 in their next releases at least as an alternative (the same way some of them shipped Apache 1 and 2). But in Horde for example we still have people running PHP 4.1.x. PHP 4.2 has been released two years ago! Don't get me wrong, I don't want to support outdated versions of whatever software forever. But I'm pretty conservative from that background.

Other valuable point I see for switching now to PHP5, is the fact that maybe more serious developers start looking at PEAR, or the fact that PEAR was always a big testsuite for latest PHP technologies.

Here it is again, the different usage of PEAR that we have. Many developers might use PEAR packages as components for a single software project on a server they control (more or less). But there's also projects like Horde that develop software being distributed widely and that depend on PEAR packages as third party components. Such projects have different needs, they can't change the environment if they want to depend on a certain feature and they can't "fix" their code if a third party component changes its behaviour. That's why I fought for BC so strongly in the past. So, what are we talking about here? We are not talking about packages that depend on new technologies like sqlite, PHP streams or any XML functionality being new to PHP 5. We are talking about about the implemenation of an algorithm, 50 years old and probably implemented in any computer language that exists. This is *not* a package that will need months to become stable. This certain package could be available to millions of webservers *now*.

Any way, IMHO, we can/should vote against a package if is missing something, doesn't conform CS, etc. but in any case if it's just written for PHP5, sounds really crazy (again imho).

If it's written for PHP 5 without reason, it doesn't sound crazy to me. Jan. -- Do you need professional PHP or Horde consulting? http://horde.org/consulting.php