Ok, i changed from reading from the database to reading from a file. It now clocks at 3.37seconds for the malignant peace of javascript.. Still too slow :( Timing the readfile statement that outputs the cache file, is at 0.00109 seconds according to a measurement done with microtime() It seems that this is not correct, because if i remove the readfile statement, things speed up to 120 milliseconds (from 3.37seconds).. Some more help would be greatly appreciated.. On Wed, Jun 4, 2008 at 9:20 AM, Nathan Nobbe <quickshiftin@xxxxxxxxx> wrote: > On Wed, Jun 4, 2008 at 3:05 AM, Rene Veerman <rene7705@xxxxxxxxx> wrote: > >> Hi.. >> >> I've built a cms that's running into a serious speedbump. >> I load several sets of javascript through a custom built db >> caching&compression system. >> Loading it without caching and compression takes between 2 and 3 seconds >> for >> the entire app, 112Kb javascript. >> >> The caching system first compresses/obfusicates the requested >> source-files, >> then gzcompress()es that and stores the result in my database. >> >> The next time the page is requested, the caching system detects the cached >> copy, and outputs that with an echo statement. >> >> One of these code bundles that i request takes over 6 full seconds to >> load. >> It's the same 122Kb obfusi-compressed to 52Kb. >> Even without obfusication, it takes 6 seconds to load when taken from the >> database. >> So it's not the browser being slow in parsing the obfusicated code. >> Firebug's "net"-tab shows this one snippet taking 6 seconds. I don't know >> exactly what that measures, just the transit time i hope.. >> >> I suspected the database query, but retrieval takes less than a >> millisecond. >> So does the 'echo' statement that outputs it. >> >> I'm really puzzled as to what can cause such a delay. >> > > i think putting the js in the database at all, and then using php to > retrieve it is unnecessary overhead. i would put the cached contents on > disc, and refer browsers to a direct url. maybe you could just put the > contents of one of your compressed files on disc, and request it from the > browser, time that, and see if it gets you anywhere. > > also, i would check into some of the resources yahoo has in this dept. for > example, the ySlow firebug plugin, > > http://developer.yahoo.com/yslow/ > there are additional resources as well, and they even have a really slick > js compressor, freely available. its so slick in fact, that it will reduce > the length of variable identifiers that are safe to do on because those > variables happen to be closures. neat stuff. > > -nathan >
