I'm currently pondering a system for storing sample files for automated and manual testing (of the UberConverter.) One goal of this is to turn this pile of files into something with some attached meta-data, and another is to allow me to easily add them to a list of files to be run through a given automated test.
Has it been done before? In a way that plays nicely with subversion? With an open source implementation? With very little setup needed on the client side?
I have not been thinking along the lines of an SQL database. Seems that would be too much hassle to get it running out of a tarball on any platform, though I guess sqlite would be an option. All the same, I'm guessing that the problems which that solves would not payoff against the burden of opacity.
So, I'm considering writing something which is largely based on convention and using subversion's builtin metadata. The propedit command allows me to drop into an editor. Also, renaming a sample file means that the info moves along with it.
But, renaming a test script leaves me needing a validator for the props. Alas. Everything is doomed to eventually be laden with formality? Feature creep at the "dreams and kittens" stage can't be helping that — I've already been picturing a webapp with wiki-like features for uploading sample files.
