Re: [Doxygen-users] Making an Easy and fast readable C Library Manual reference.(TODO-Suggestions)

[Carlo W. <ca...@al...>]

On Wed, Nov 21, 2001 at 08:30:31AM -0000, Stephen Goudge wrote:
> might even _want_ to use it) and I'd feel a fool if I prevented that
> because I'd been precious about my wonderful design :-)

We're clearly not talking about the same.
I tried to clearify that in a previous post, but failed apparently.

All of my header files are user-level and all of them are needed
by the user.  All of them are documented.  But just ONE needs to
be included, because it includes all others.

The "solution" of putting them in a separate directory simply
doesn't work.  This has nothing to do with doxygen *forcing* me
to use a better design of my library; instead, doxygen is simply
failing to deal with this kind of design and it would be a whole
lot better when it was fixed to NOT speak about header files
that are in the INPUT list but do NOT contain a \file command.
Evenmore, it should never print a #include <suchafile.h> when
it can see that this file is being included by a *documented*
header file.

In other words:

header1.h  : contains a \file entry and #include <header2.h>
header2.h  : does not contain a \file entry, but does define
             a documented function.
The function of header2.h then should be documented with an
#include <header1.h>, and there should not be any mention of
header2.h anywhere.

If someone wants header2.h to be mentioned in the File List,
then they should add a \file entry.  If then still they would
like to see a #include <header1.h> then they'd add an explicit
command to header2.h to achieve that (if at all possible).

I hope you can see the logic in this.

-- 
Carlo Wood <ca...@al...>