Derek recently blogged that “phar is cool!” Cool is great … but it doesn’t answer important questions: how does loading your application from a .phar file affect overall performance and scalability?
How well does it work with leading bytecode caches?
Where are the benchmarks for phar? 