[qmtest] bug in file_database.FileDatabase

[Stefan Seefeld]

Mark Mitchell wrote:

> The reason for labels is that it is reasonable to map tests into storage
> domains other than the file system, like, say, a relational database. 
> Those domains may have different rules about how you spell paths and
> such than the UNIX file system.

I agree about the differing needs. But why should qmtest care about them ?
If some users have to bind qmtest to tools with incompatible requirements
as to test naming conventions, they can always use a little adapter.
But is that a reason to add another abstraction (with the implied complexity)
*into* qmtest ?

> Nathaniel and Zack and I talked about this a lot this past summer, and
> couldn't really come up with a cleaner model.

Ok, no need to reopen this thread then. Still it would be good to have
the rationales documented somewhere so people can look them up before
making enhancement requests :-)

Kind regards,
		Stefan