[Clipart] Searching and download graphics (3)
oneminuteinspirations at gmail.com
Thu May 17 13:50:25 PDT 2007
You don't gain any significant efficiency by grepping text files over
storing the same information within a database. As well, the search
utility offered on the OCAL site was primarily text based. Though they
would sometimes display a PNG, more often than not it was just text.
Additionally, the server load to serve a 50k .gif is the same as to
serve a 50k text file. Graphics, unless being dynamically generated,
only eat up bandwidth in the pipe, but don't really tax the server.
It's the database queries and script processing that are the problem.
The ways to gain efficiencies there are to pre-make the output of
certain popular database queries and update them periodically (such as a
tag cloud that updates every 6 hours instead of updating live), or
offloading some of the search effort to the client (via a Flash app that
downloads an XML file and builds its own search index from it) or a
third party (such as Google).
> Basically, the search and show utility would be primarily text based,
> which should allow for extremely rapid searches. The graphics would
> only be shown after the search had completed. The graphics shown would
> be based on the unique key(s) brought up by the search.
> Again, I don't have anywhere near the technical background of a lot of
> you, so you would need to tell me if this can work or not.
> To keep the text-based search field updates low, you could organize
> based on month. All submissions are organized based on date into text
> files organized by month and year. The search fields and download
> locations could be incorporated into these same text files and easily
> clipart mailing list
> clipart at lists.freedesktop.org
More information about the clipart