Re: PR11067 patch

[Joel Brobecker <brobecker@adacore.com>]

> But what's the cost?  Functionally, it consists mostly of replacing
> an fputs_filtered with a vfprintf_filtered, so the performance hit
> is negligible, and the patch adds maybe a hundred lines of code,
> most of it in setting up the set enum-fmt command and translating
> the format strings.

One lesson we (AdaCore) learnt from developping GPS (AdaCore's IDE), is
that adding options and flexibility can quickly cause the amount of testing
to explode.

I agree with Tom in this case that it's better to not have the formatting
option.  I know it can be discouraging sometimes that everyone bikesheds
a bit on things that small.  I actually think it's a good thing that
people provide their opinions, even at the risk of mild bikeshedding.
I'd rather have all opinions now rather than later.

But there comes a point when we need to make a decision, and I think
we have reached that point (or went past it? ;-) ).  I personally like
Pedro's suggestion, but I'm ready to accept any format at all. I don't
think anyone really expressed a strong disagreement, so I'd just review
the thread, select the format that seemed to be prefered based on feedback,
and just announce that this is what you'll implement.  I'm sure some will
feel like they must propose something that's obviously nicer than your
proposal, but if there are no objection, I suggest we go with what you
announced in the end.

-- 
Joel