Re: Re: mod_perl/Apache question.

...you can always write your application as an Apache::Registry app (Apache::Registry is a tool for running CGI scripts through mod_perl), and then it can be portable between the two systems without any changes...

In my opinion, you don't even have to do that. The excellent Apache::compat module mimics most of the 1.0 API on mod_perl 2.0. I've yet to have a customer that couldn't migrate without problems from 1.0 to 2.0 using that module.

Liz

Comment on Re: Re: mod_perl/Apache question.

Replies are listed 'Best First'.
Re: Re: Re: mod_perl/Apache question. by perrin (Chancellor) on Oct 15, 2003 at 15:35 UTC
That's true, but using Apache::compat is somewhat slow because of the extra work it does to emulate mod_perl 1 functions. It's meant as a temporary stopgap to port old code to mp 2. UPDATE: I asked for opinions about this on the mod_perl list, and the answer I got was that Apache::compat is problematic because it makes some code written to the actual mod_perl 2 API not work. Also, it uses a bit more memory than Registry because it loads a whole bunch of optional modules. Actual performance comparisons are hard to draw though, since it depends on which functions you use. Some functions simulated in Apache::compat are going to be significantly slower than they were in mod_perl 1, but not all of them. The full response is here.	[reply]

Replies are listed 'Best First'.

Re: Re: Re: mod_perl/Apache question.
by perrin (Chancellor) on Oct 15, 2003 at 15:35 UTC

UPDATE: I asked for opinions about this on the mod_perl list, and the answer I got was that Apache::compat is problematic because it makes some code written to the actual mod_perl 2 API not work. Also, it uses a bit more memory than Registry because it loads a whole bunch of optional modules. Actual performance comparisons are hard to draw though, since it depends on which functions you use. Some functions simulated in Apache::compat are going to be significantly slower than they were in mod_perl 1, but not all of them.

The full response is here.

[reply]