Maintaining meaningful directory layouts

[marq]

I have realised that it has become practice to put data files for specific tests into code subdirectories, and maybe even test scripts into the same source code directories.

The orginal layout of the directory structure for all ROPP packates was that we always have several directories for containing the pure code The contents of these directories is what the user requires to run / operate the software. data files and test scripts do not belong to this category.

There tends to be a tests subdirectory which contains tests for the user in order to check if the compiled code works reasonably well. Full scale testing should be done in ropp_tests, but nor become part of the distribution. For data files, usually a data subdirectory exists. This - that was the idea - contains either sample data (which may also be used for user tests) or some setup data that the user may choose to use - or not.

By mixing these categories we make the directory structure more difficult to maintain and understand, so I would appreciate if we stick with the original layout ideas.