These are a few notes I’ve taken during the presentations I attended at the 2007 edition of the London Perl Workshop.
Scaling with memcached (Leon Brocard [acme])
- Don’t underestimate the fact that this is a cache, not a database –can’t dump, can’t iterate over keys, not persistant, not redundant, etc.
- TTL is good, invalidation is bad (which means things will get out of date)
- Tuning is done with TTL, cache sizes
- Uses memory, not so much CPU
- Easily distributed (over servers, CPUs, etc)
- Limits set by default are small, change them!
Catalyst quick start (Matt Trout)
- Not a “do it fast” approach to stuff, a bit more work to get started, but supposedly better at tackling your problem
cpan Catalyst::Devel
- App scaffolding:
catalyst.pl Project
- very basic…
Catalyst is basically just a component loader and a URI dispatcher
- UTF8 out-of-the-box (
DBIx::Class
may be helpful)
Controllers (URI dispatching)
:Local
,:Global
,:Regex
,:LocalRegex
:Chained
-> powerful enough to implement almost everything else…
Models
- DBIx
- Not the point of this tutorial, JFGI
Views
- Template Toolkit is widely used
HTML::Mason
(ouch!)HTML::Template
(too simple for most stuff?)
Deployment
- mod_perl is OK, but probably overkill
- FastCGI is perfect for new stuff that only uses Catalyst