Suggestions for GNATS

[Neil Booth <neil at daikokuya dot demon dot co dot uk>]

I've got a couple of suggestions to help improve the quality of bug
categorisation as reports are entered into GNATS.

One of my complaints is that a lot of reports get dumped into the
catch-all categories of "c" and "c++" for no reason other than the code
being compiled is c / c++.  I'd like to rename these "C front end" and
"C++ front end" respectively, to make (what I assume) is the intended
meaning more clear.

I think there is good cause for a category "bootstrap" or "build"; I
prefer the former as its intention is clearer.  Currently most bootstrap
bugs are placed by submitters in the "c" category.  It would also make
it easier to close these - people who've bootstrapped on XYZ host or
ABC cross can quickly lookup any bug reports for those and close them.
I also think we should discourage input of bootstrap bugs; most tend
to be short-term breakage in the tree that everyone knows about, but
it's hard to reproduce the submitter's exact conditions later to know
the report is safe to close.  Perhaps we should advise that they not be
entered unless the inability to bootstrap has lasted for several days,
or if the submitter believes his target to be not used by the main GCC
people.

I think it might be an idea to have a "don't know" category - I think
it's better that submitters who aren't sure of the right category say
so explicitly than dumping it in the wrong one - I'd be happy to tidy
up after them.

It might also be useful to have a new field labelled "hardware" or some
such, consisting of a drop down limited to the common platforms for
consistency of description.  At present I can't tell if I'm going to
be able to test a bug report until I open it and look at the hardware
specs.

Finally, I'm happy to write a one-or-two paragraph "rough guide" to
explain the differences between the categories and how to tell which
one to put a report into.  I'd say roughly 40% of submitted reports
are put in the wrong category.  Just lowering that to 20% would save
everyone perusing the reports a lot of time.

Neil.