Re: [PEPr] -1 for Authentication::NTLMProxy

On 7/4/06, Pierre <pierre.php@gmail.com> wrote:

On 4 Jul 2006 19:41:49 -0000 papercrane@reversefold.com ("Justin Patrin") wrote: Justin Patrin (http://pear.php.net/user/justinpatrin) has voted -1 on the proposal for Authentication::NTLMProxy. Proposal information: http://pear.php.net/pepr/pepr-proposal-show.php?id=396 Vote information: http://pear.php.net/pepr/pepr-vote-show.php?id=396&handle=justinpatrin Comment: I'm sorry, I can't approve this package. You didn't even try to use HTTP_Request, even after I spent the time to implement Keep-Alive. Despite what Pierre thinks, PEAR packages should not reimplement the same code without a very good reason. A good reason was not given for this package, especially since HTTP_Request can now make use of Keep-Alive with the patch I wrote. You need three good reasons? Speed, maintenance, easiness.

*bzzzz* Sorry, try again. It may be somewhat faster, but reimplementing code is not acceptible in this case. Maintenance is decreased when you *don't duplicate code*. One place for errors to happen, one place to update to fix *everyone's* problems. The only valid reason for not using HTTP_Request was lack of Keep-Alive support. I implemented this and it was ignored. Easiness is also a false argument in this case. using HTTP_Request internally will not increase the complexity of the external interfaces of this class and could even make it more consistent with other PEAR packages. In fact, using HTTP_Request will actually *reduce* complexity by making the code for this package simpler and more easy to understand and make maintenance of this package *easier* by reducing its lines of code.

It is plain wrong to consider to use HTTP_Request (which uses Net_Socket (*g*) and) for such needs. How anyone can seriously ask to use it for a NTLM handshake? It is ridiculous.

No, it is not ridiculous. One of the basic tenets of PEAR is code reuse. To reimplement HTTP code is what is ridiculous. If you have a problem with HTTP_Request and Net_Socket try contributing some patches to make them more efficient instead of trolling.

About die and exit, you should at least *read* the updated source code before voting.

Ah, I see die() is gone. however, exit() is not. The concerns were not addressed. See previous comments and e-mails.

Also consider to stop your personal vendetta, many people has the same opinion on this topic.

I am entitled to my own opinion and my own vote. -- Justin Patrin