From: David Abrahams (david.abrahams_at_[hidden])
Date: 2002-07-21 06:53:33
There's a time for worrying about what's in the standard, and there's a time
for looking at what's going to work well in the contexts in which Boost is
deployed. At this point I think it almost doesn't matter what the standard
says about headers, especially since it says so little with any definitive
clarity. I think Boost should do what other "similar" libraries do, whatever
that is. We ought to look at how things like tools.h++, QT, etc. get used.
"Gennaro Prota" <gennaro_prota_at_[hidden]> wrote in message
> On Sun, 21 Jul 2002 00:50:17 -0700, "Victor A. Wagner, Jr."
> <vawjr_at_[hidden]> wrote:
> >Well, Stroustrup (3rd edition), Section 9.2.1 (yeah, I know, not the
> >refers to "header files" as both the ones the system has, and the ones a
> >programmer writes so that definitions of objects, functions, classes are
> >consist ant.
> That's the usual terminology, of course. And it could be the case that
> Mr. Stroustrup is simply adhering to it (just like he talks of
> 'preprocessor' and also of 'preprocessor', 'compiler' and 'linker' as
> three separate things). It's obvious that he knows very well the
> details of the standard but his book has also (see the preface) a
> certain regard towards tutorial aspects.
> In any case what I don't like about similar terminologies (saying that
> anything you #include is a header) is that they lose the connection
> with the entity to which they refer, i.e. (I hope to be clear, I'm not
> a native english speaker) that one should say: the header OF A
> #include is the header identified by ITS the h-char-sequence part.
> In other words, there's nothing that is a header PER SE, just like
> there's nothing that is an addendum or a factor per se, and the same
> number can be either one or the other. More strictly, when you say
> that for instance 3 is an addendum, you are actually making an abuse
> of language to say: "I'm considering the value of the addition
> corresponding to a couple of numbers (3, x) or (x, 3)". Analogously,
> with that terminology, saying "a header" would simply mean "a
> #included thing".
> Unsubscribe & other changes:
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk