Sure, libapreq and its included modules are the fastest way to parse
parameters and handle cookies. However, hardly anyone needs that
speed. Most people are spending all of their time talking to a
database or doing other I/O tasks, or have pretty minimal web traffic,
or both.
That's true. I spend more time optimising the database and the web ui
then anything.
If you're writing something new, and you don't care about supporting
CGI, and you want to use the fastest option, by all means give
libapreq a try. For most people though, it won't make any measurable
difference which one you use. I wouldn't advise anyone to bother
changing CGI.pm code unless they've run a profiler and seen it show up
high enough to matter.
Apache2::Request, and Apache2::Cookie looks like compatible replacements
to CGI and CGI::Cookie. I think that's the objective anyway, so that
when the need arises, it's just a simple replacement from the root
function call (assuming factoring has been done). I think the difference
will be whether the web application is a high traffic site-low cpu site,
or a high cpu-low traffic one. The performance gain for the former will
be more significant.
