Re: feb-2002 libstdc++ outstanding

[Paolo Carlini <pcarlini at unitus dot it>]

Benjamin Kosnik wrote:

> 2002-02-18 Re: [PATCH] Fix libstdc++/5708
>         http://gcc.gnu.org/ml/libstdc++/2002-02/msg00148.html
>         http://gcc.gnu.org/ml/libstdc++/2002-02/msg00161.html
>         http://gcc.gnu.org/ml/libstdc++/2002-02/msg00163.html
>         http://gcc.gnu.org/ml/libstdc++/2002-02/msg00181.html
>         http://gcc.gnu.org/ml/libstdc++/2002-02/msg00321.html
>         a confusing mess that nobody understands except for Nathan?
>         money_put and what 'space' means, exactly.

;-)

My personal opinion is that, right now (after
http://gcc.gnu.org/ml/libstdc++/2002-02/msg00148.html) we have a behaviour more
consistent than before but *not* a strictly conforming one. We have therefore at
least two options:

1- Stick with the current behaviour not strictly conforming but similar to that
of Metro and Rogue and wait for a rewording.
2- Implement something *simple* but 100% conforming according to our current best
interpretation (for instance we choose to output an ' ' for the space field,
always *followed* by fill chars, that is something like
http://gcc.gnu.org/ml/libstdc++/2002-02/msg00167.html)

At the moment, with the standard very vague on those matters, I don't think is an
option implementing a sophisticated solution with ' ' either following or
preceding the fill chars depending on the pattern.

Opinions to the contrary?

Ciao, Paolo.