From: Mark Rodgers (mark.rodgers_at_[hidden])
Date: 2000-06-01 06:18:33
From: Dave Abrahams <abrahams_at_[hidden]
> on 5/31/00 8:55 PM, jsiek_at_[hidden] at jsiek_at_[hidden] wrote:
> > 4. Do what STLport does, and bring those names into std if necessary.
> 1. That's a privilege reserved to implementations (when you use the
> you are changing implementations).
If we have a non-conformant implementation, I'm not sure what harm there
is in doing something non-conformant (for that implementation only), that
fixes the non-conformance. So although I may agree with you, it is not
because of the point you are making here.
> 2. I don't want to get into the business of trying to make non-conforming
> compilers look conforming to code outside boost. That really seems
> our charter.
Agreed. I would favour writing conformant code and putting std::ptrdiff_t
when that is what you mean. If this doesn't work with a particular
implementation then you can document that problem and suggest possible
workarounds in implementation specific documentation. Some users may be
happy to bring the names into the std namespace themselves, others may
prefer to change library vendors, and others might prefer not to use your
code. It should really be up to them.
What I don't really want to do, is pay the price of having other users'
non-conformant implementations affect the readability of the code I write
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk