Hi,
__declspec() is not only a borland extension. MS MFC and ATL libs also =
use
this stuff. You can work around this problem with help of the =
PREDEFINED tag
in a configuration file. In my configuration file I have the =
MACRO_EXPANSION
tag set to yes and a long list for PREDEFINED. You can handle =
__declspec
like a macro. Try something like this:
PREDEFINED =3D "__declspec(mod)=3Dwhatever you want to have in your
documentation"
I don=B4t expand __declspec() so my entry for this is:
PREDEFINED =3D "__declspec(mod)=3D "
With that, the __declspec() will be ignored by doxygen (to be correct, =
it
will be replaced with an empty string I think). If you want to =
interpret it,
mod contains your modifier (for example dllexport). I think this should
work.=20
Take a look on the Preprocessing section of the doxygen documentation, =
there
it is described with some examples.
hope this helps,
Aaron
Message: 6
Date: Fri, 18 Jan 2002 15:26:56 +0100
From: Hofi <ho...@fw...>
To: Doxygen users list <Dox...@li...>
Subject: [Doxygen-users] once again about BCB modifier keyword =
extensions
Hi All,
Last time i wrote about problems with __declspec(dllimport) and
namespaces.
I found that all modifiers like __declspec(<decl-modifier>)
where decl-modifier could be
dllexport
dllimport
naked
noreturn
nothrow
novtable
property
selectany
thread
uuid
could confuse doxygen :((
classes with __declspec() are all ignored,
variables like
//< Initialization of the default global instance of a logger list.
HLoggers __declspec(dllexport) loggers;
are miss interpretered.
e.g.
const char HLib::__declspec ( dllimport ) [inline]
i think the reason doxygen works this way is because __declspec is a =
borland
language extension only :((
so i only can hope that doxygen could interpret these modifiers in the
future, it would be a nice thing from Dimitri.
Thanks
and sorry for my terrible english.
bye
Hofi
|