Re: [SMARTY-DEV] Re: speed improvement idea: path caching

Sorry if this is real obvious, but another cool thing about having a path cache per called template is how it can be used to identify the need to compile--especially if timestamps are included with the path cache and if all path cache files are stored in a single directory. 1) path cache file does not exist? - compile template - write new path cache file 2) path cache file exists && no compile check? - only load compiled template (don't load path cache file) - (is that true? I don't see why it would be needed) 3) path cache file exists && (compile check || force compile) - load the path cache file, iterate and match timestamps compiling as required. write new path cache file(s) as required. If it is assumed that a compiled file MUST exist in the case that its corresponding path cache file exists then additional assumptions can be made: eg. the compiled file is readable, the compiled file is writable, etc. This means that the extra file system call overhead of using an additional file per called template can potentially be offset by reducing the number of filesystem calls currently made on compiled templates. Of course, the first time you compile, you'd still have to make appropriate checks, but after that, you can can relax a bit and rely on the cache path files alone to determine compilation needs. a crazy idea: write the cache path files to shared memory using something like pear::Stream__SHM on compliant systems. Again, sorry if that was obvious but I thought it was really neat. I'm sure Monte has something far more clever in mind. :) xo boots __________________________________ Do you Yahoo!? SBC Yahoo! DSL - Now only $29.95 per month! http://sbc.yahoo.com