|[4 earlier articles]|
|Re: `javadoc` for c++ firstname.lastname@example.org (1998-10-17)|
|Re: `javadoc` for c++ Marko.Makela@HUT.FI (Marko =?ISO-8859-1?Q?M=E4kel=E4?=) (1998-10-17)|
|Re: `javadoc` for c++ email@example.com (1998-10-17)|
|Re: `javadoc` for c++ firstname.lastname@example.org (Bruce Dawson) (1998-10-17)|
|Re: `javadoc` for c++ email@example.com (Jerry Leichter) (1998-10-17)|
|Re: `javadoc` for c++ firstname.lastname@example.org (John Carter) (1998-10-22)|
|Re: `javadoc` for c++ email@example.com (Matt Timmermans) (1998-10-30)|
|From:||"Matt Timmermans" <firstname.lastname@example.org>|
|Date:||30 Oct 1998 13:08:32 -0500|
>On Tue, 13 Oct 1998, Marcus Rohrmoser wrote:
>Mix the C++ function declarations, documentation and implementation in
>one file. Mark it up in XML. (See www.w3.org of info on XML)
>Run it through James Clark's nsgmls to validate and parse the
>XML. Feed that output into a wee perl scriptie from which flows a whole
>bunch of streams.
>1) The class interface (.h file)
>2) The class implementation (.cc file)
>3) The online documentation. (.html file)
>4) The dead tree documentation (.tex file)
Uhhmm.... And what is the C++ for, exactly? Wouldn't we just call
the XML language CodeML and compile it to executable code?
This sort of thing pops up every now and then in SGMLland, and it's
almost always a bad idea. The syntax of a programming language like
C++ is designed to statisfy many requirements, including efficiency,
expressivity, familiarity, readability, and applicability to the
problem domain. Pick a function... quicksort, say... and write it in
CodeML (remember that you have to be able to extract adequate
documentation from this), and it will be clear why no SGML processing
language uses an SGML-based syntax.
Return to the
Search the comp.compilers archives again.