Login | Register
My pages Projects Community openCollabNet

propel
Reply to message

* = Required fields
* Subject
* Body
Attachments
Send reply to
Topic
Author (directly in email)
Please type the letters in the image above.

Original message

Author Soenke Ruempler <soenke@ruempler.eu>
Full name Soenke Ruempler <soenke@ruempler.eu>
Date 2006-10-03 14:05:28 PDT
Message On 03.10.2006 21:50, Scott Wehrenberg wrote:

> On 10/3/06, Hans Lellelid <hans at velum dot net> wrote:
>> Another option would be to write a Phing task that builds all dependency
>> classes into a single file. This doesn't seem that difficult -- and
>> it's what Phing is great at. Then users can "build out" their Propel
>> package for performance reasons if they want to have everything packaged
>> in a single file. I don't see why this would be incompatible with other
>> ideas proposed.
>>
>
> I would love to see some speed comparisons between the single big file
> approach and the autoload method that's being worked on now. I'm
> guessing that the single big file would lose out without an opcache
> except in the rare circumstance where you're working with basically
> all of the classes for a request. It'd be a little more interesting to
> see how each did with the opcache.


Even with an opcode cache you have to respect the sandbox-principle of
PHP. The Class/Function/Object-model is built upon every new request.
Opcode Caches just eliminate parsing and compiling, but every
class/function must be registered in the engine though.

So I'm wondering if there are performance improvements or even losings
in a big object model with one big file. Maybe (IMHO) just using (and
registering) the classes that will be needed is faster!

-soenke