Re: egcs 1.2 C++ ABI (Re: C++: STL 3.2)

["Ross Smith" <ross.s@ihug.co.nz>]

From: Jason Merrill <jason@cygnus.com>
>
> >> But it is backward compatible.  I don't mind requiring people to use
> >> the newest compiler to link mixed code.
>
> > Ok. So 1.2 is backwards-compatible in the sense that you can link 1.1
> > libraries into 1.2 applications, right? (Unless they use STL).
>
>Right.

I'm not sure how much effort is being put into this issue, but there's
a question of perspective that you may wish to keep in mind. (I'm
offering it just as a suggestion. Not being a direct contributor to
EGCS, I'm under no illusions about having any right to an opinion :-) )

A lot of C++ programmers who learned the language in the days when the
ARM or Cfront was the definitive standard still tend to think of the
STL as something new and exotic, that only needs to be brought in
when something really advanced is needed. But those of us who have
picked up the new concepts that come with standard C++ tend to use a
different style, in which the STL is all-pervasive. I can't imagine
writing more than a handful of lines of C++, and certainly not even
the most trivial complete program, without using the STL.

What I'm saying is that, if you're going to the trouble of making EGCS
"backwards-compatible ... unless they use the STL", you may want to
reconsider whether it's worth the effort, because for anyone who uses
a modern style of C++, "it works unless you use the STL" is for all
practical purposes synonymous with "it doesn't work". If you need to
break compatibility on the STL (and I gather you do at this point),
you might as well take the opportunity to break anything else you have
an issue with as well.

--
Ross Smith ................................... mailto:ross.s@ihug.co.nz
.............. The Internet Group, Auckland, New Zealand ..............
    "Perl is the Unix way. 500 million ways of doing the same thing,
    and 500 million monster egos all insisting on their way being
    the Proper way of doing it." -- David Parsons