[Xitami] lrwp php?

[Paul Reid]

> searches through some large text files, I'd like to have the text files loaded in memory so they don't keep being read from disk during each query.

How large? Are you sure they are actually being read from disk? 

In unix and Windows, most available memory is used as disk cache. If you
have enough RAM to avoid thrashing, the second third etc time you read a
file that hasn't changed on disk, it usually comes from cache rather
than from physical disk.

You could test this by adding code to grab the time when the program
starts and is done, compute the difference, and write that at the bottom
of the search report. If I'm right, the very first time each day (or
re-boot, or RAM-overload) that it runs will be seconds, each subsequent
search will be milliseconds.

You might also grab the time before and after the file read and before
and after the search process, if they are two phases. In VB, a
case-sensitive search turned out to be FAR slower than any disk-work. 

I'm also sure that for VB and Perl, the interpreter start-up takes
longer than about any code I can write in a month. 

LRWP is a good idea, but hard to get working. Be sure you need the speed
and that it will really help before you go to that extreme. Many massive
sites work fine with plain CGI. 

 -PRR