tags:

views:

33

answers:

1

If you have a C++ project that has a bunch of .ODL files and the generated .h files from the ODL compiler, should doxygen be told to parse both .odl and .h, or only one or the other? In general I don't like documenting generated code but IDL is sort of a special case.

In any case, it seems like the member listing of ODL files is not quite working properly in my tests, are ODL files properly parsed?

+1  A: 

From this page it seems the Doxygen perfectly supports parsing *idl and *.odl files. Maybe you have some issues regarding the comments format ?

Anyway, I would advise you against parsing both *.idl/*.odl files and generated .h files. The latter are, by definition, generated and represent the same classes than the one defined in the *.idl/*.old files.

Parsing both is like documenting the same set of classes twice, which is obviously not a good idea.

ereOn
That was my initial thought... but what happens with classes using these interfaces... they include the .h files so wouldn't that mean a 'gap' in the chain?
John
I'm not sure about the comment format, I have it set to generate everything even if there are no source-comments (we're doxygenating an existing project). I get a link to "xxx.odl" but clicking on it doesn't show me the file's source like with .h files.
John
@John: I doubt Doxygen can generate links to generated files (aka. files it did **not** parse itself). About the "gap" in the chain: i guess it's just a way of seeing things ;) Imo, it is perfectly fine to think that some not-generated classes use generated files. There is no real "gap" here: just an additional "link" (the "generation").
ereOn