On 6/7/07, Tina Müller <[EMAIL PROTECTED]> wrote:
but you still save CPU if you have fast templating. maybe you can save a number of machines.
Unless your templating engine is showing up pretty high on your Devel::DProf output, you're probably not going to save much. Saving 5% won't mean saving a machine for anyone except really large sites with very well-tuned machines.
when i said that many people don't care about speed because of their database slowness i also meant that i wonder why people don't rethink their architecture. if the database is so slow that it makes no difference how slow your perl-code is, then there's often a possibility to optimize that (cache things, use a search engine for searches instead of the database, change replication architecture, ...)
I agree, and that's where people who are seeing database activity as a bottleneck should spend their time. They can save time on their templating engine choice by choosing the one with the development features that they find most useful and not trying to figure out which one is fastest. All I'm saying is use your optimization time wisely. Good error messages and an easy API are more important than speed in a templating tool now. If the dot notation in HTML::Template::Compiled saves a developer a few hours of coding which can then be spent on database query tuning, that's probably a much bigger win than any speed difference.in templating modules. - Perrin
