A bug in the current _M_insert_float/_M_group_float

[Paolo Carlini]

Jerry Quinn wrote:

> The problem is definitely still present in my version of floating
> point. I didn't touch grouping processing in my stuff.

Indeed. But I wanted to inform you ASAP as the person actively working
in this area.

> Issue 282 says adds floating point grouping. Issue 358 says don't
> group beyond the decimal point. Perhaps another issue is needed to
> say that grouping shouldn't be applied to the exponent section. Or
> could issue 358 be extended to deal with this as well?

I think so. In any case, whatever we chose to do in an area where the
standard it's not completely clear (but almost, I would say, after 282
and 358!), it must be something meaningful: definitely it's not a good
idea feeding __add_grouping with 'e'/'E' and the exponent sign and
everything!

Paolo.