Re: Standard header format.

[Matt Austern <austern at apple dot com>]

On Jun 5, 2004, at 10:30 PM, Steven T. Hatton wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Sunday 06 June 2004 00:25, Gabriel Dos Reis wrote:
> > "Steven T. Hatton" <hattons@globalsymmetry.com> writes:
> > I have never suggested he didn't know what he was
> > | talking about.
> >
> > You did:
> >
> > #                                                     I believe 
> > Stroustrup
> > was # operating under the traditional assumption that the standard 
> > headers
> > are in # fact header files.  I do believe that's what he meant.
>
> My intention was that he assumed the implementation the reader would 
> most
> likely encounter would have traditional header files. That's why I 
> qualified
> 'assumption' with 'traditional'.
>
> There are many C++ programmers who have suggested to me the best way 
> to learn
> what the standard headers probide is to look a them.

Irrespective of whether Bjarne thinks or thought that that was a good 
idea, I don't.  A production version of the standard C++ library will 
use all sorts of tricks, for all sorts of reasons.  Some of the design 
goals for a production implementation are: standard conformance (which 
requires "uglifying" symbols and breaking circular dependences), 
support for legacy code that used pre-standard versions of the headers, 
improving run-time performance, improving error detection, improving 
compilation speed, reducing memory usage.  Implementers will use 
whatever tricks they have to to achieve those (and other) goals, some 
of which use compiler magic and some of which make the source code less 
transparent, from the perspective of someone who isn't already very 
familiar with the library, than it would otherwise be.  A production 
implementation of the standard C++ library is unlikely to be a good 
pedagogical introduction.  Anyone who expects it to be one is going to 
be disappointed.

			--Matt