I know that ZF strives to provide a balance between powerful tools and easy
of use. However, reading this thread and others like it, it seems to me that
there are two approaches "battling" here ;) The first one is the one
concerning small to medium, shared hosting applications, while the other is
for the considerably larger projects, projects with dedicated (or more
sophisticated) hosting. In this thread clearly we can see the positions:

* shared-hosting ZF projects are expected to be harmed by "hardcoded"
includes, because having all those exception files loaded is not necessary
and seems like overhead; those are projects that run in environment with no
opcode cache solution available (as more or at least some of the shared
hosting providers).

* ZF projects flexing opcode caching solutions are expected to me harmed by
"conditional" includes, because they will not be picked up by the opcode
caching.

I think there are other areas in ZF like this as well. What I propose is to
consider forking ZF into 2 "versions" - one that suites best the
shared-hosting projects (small and medium projects) and one for the
dedicated projects (large, enterprise projects). It is probably too early
for a split like that, but at least it is worth discussing it: this seems
like a very sane choice, because each "version" will be true to the purpose
it is going to be used for. The API will be the same, just the
implementation of certain parts will differ slightly to fit best the
environment the library will be used in.

What do you think ?

-----
K.


Regards,

Kaloyan K. Tsvetkov

http://kaloyan.info/blog/ http://kaloyan.info/blog/ 
-- 
View this message in context: 
http://www.nabble.com/Reducing-the-number-of-loaded-exception-files-tp14324215s16154p14566139.html
Sent from the Zend Framework mailing list archive at Nabble.com.

Reply via email to