John Brown said:
I was also told by another committee member sometime ago (his name
escapes me off-hand) that members even have an HTML version but the ISO
itself is responsible for its release as a ".pdf" file.
There is indeed a HTML version used by committee members. However, this
version is not always 100% accurate and is occasionally missing some stuff
or has mangled text. Thus, it is not fit as a standard document: the
authorative version is the PDF one, even though the HTML is often used
when working on this stuff.
The document may be
just an electronic copy of the final standard but I suspect many developers
(myself included) frequently reference it. Proper page numbers, links, and
the ability to move back and forth to previously visited pages like in one's
favorite (HTML) browser would be a welcome relief.
If it so much of a relief, you might want to consider joining the committee
and get the [unofficial] HTML version...
I could contact the ISO
if they're ultimately responsible but I seriously doubt it would go
anywhere. An official source would carry much more weight. Your comments?
ISO is not responsible for releasing documents to the public. This is the
job of the national bodies, eg. ANSI, BSI, or DIN. Since these bodies
receive a reasonable part of their income from selling standard documents
it took already a huge amount of arguing with them to offer any electronic
version at a reasonable price at all (my understanding is that C++ was the
first standard made available in electronic form at low costs thanks to
Tom Plum pressing the issue with ANSI; recently BSI agreed to publishing
the C and C++ standards in form of a book at relatively low costs thanks
to Francis Glassborow pressing the issue with BSI). I doubt that any
standards body would agree to an easily modifyable format, like HTML: the
effects of circulating modified versions (well, of course, both circulating
and modifying version would be illegal) would be pretty bad! Actually,
some people still use old HTML version of CD1 or CD2 which bad enough
already...
Another complication is the production of the HTML in the first place: the
standard document is still maintained in troff. Actually, I think it is
even a modified version of troff which means that even if there is a version
of troff (or groff) which would be capable of creating HTML, it would not
be able to process the sources for the C++ standard (the reasons for this
lie in the history of the standard). There was some discussion of moving to
a more modern source but this isn't an easy task: apart from the effort
needed to convert the sources, there are some neat features of the current
system which are not readily available in alternatives. The HTML used
internally by the committee has some flaws which isn't that harmful because
the people working on the respective section normally know what should be
there.
BTW: there was no standard voted on in 2003. The current C++ standard is
still the one from 1998. The 2003 document is the 1998 standard augmented
with the changes from 2003 Technical Corrigendum. However, the resulting
document is not really binding although implementers are likely to resolve
conflicts and ambiguities in the 1998 standard to follow the Technical
Corrigendum. The next C++ standard will not become finished before 2007.
Actually, I personally doubt that it will become finished before 2009
barely making the C++0X goal...