Possible levels of diagnostics in ROPP

[Ian Culverwell]

User Kjartan Kinch has raised the following issue (in connection to ticket #255):

Finally this lead us here at DMI to a little bit of a discussion
about levels of diagnostic. We will probably want to be adding
variables to diagnostic output now and then and this might become a
hassle for you - particularly since I've just asked you to expand
the documentation of diagnostic variables. We talked about whether
there might be a sense in having three levels of diagnostic output.
Something like: NONE, STANDARD and FULL. Where NONE would
correspond to the current setting diag = .false. .  STANDARD would
then be a limited, but fully documented, set and FULL would be a
longer list that we could add to without causing too much pain and
that was primarily intended for internal use.



My own personal view is that right now such an approach is probably
overkill and we are fine with diag = .true./.false. right now, but
you can turn the idea over in your head.

Certainly sounds feasible. Would it be useful? Opinions?