On Tue, May 22, 2007 at 01:31:53PM -0500, Karl Berry wrote: > creating a cronjob that creates an index.html containing the listing > > What? Isn't any time simply a matter of how long it takes all those > bytes to transfer over the net?
I was thinking about what read operatings are needed to fetch the 'directory' listing from the HDD. The speed of that operation is dependant of the underlying filesystem, but from my understanding of filesystems this is likely to cause quite a lot of read operations on the device. Having a static file there are only read operations for reading that one particular file involved, but none for reading the directory structure (I know, it's only one level), 'finding' filenames, reading metadata (do we really need to inform users about when a directory has been last modified?). Then the output 'file' needs to be generated too, which is likely to consume some CPU time. Having a look at how long it takes to load the page (around 10 seconds here) and the size of the generated page (around 460KiB) and then applying some basic math it takes a lot longer than downloading a file of roughly the same size from dl.sv.gnu.org (491KiB with 200KiB/s). That is why I suspect that having a static file could (and probably would) increase speed and decrease the load when performing that operation. > This is why I suggested having a short index.html with a link to the > long listing. That sounds like a good idea to me. Still, we could still generate a static long listing file. Regards, Stephan
