Re: linkification warnings

[Stefan Kost <ensonic hora-obscura de>]

Am 19.12.2009 13:11, schrieb David Nečas:
> On Fri, Dec 18, 2009 at 07:42:00PM +0100, Nicola Fontana wrote:
>> Il giorno Fri, 18 Dec 2009 16:24:28 +0100
>> David Nečas <yeti physics muni cz> ha scritto:
>>
>>> On Fri, Dec 18, 2009 at 02:58:10PM +0100, Nicola Fontana wrote:
>>>> Il giorno Fri, 18 Dec 2009 13:18:02 +0100
>>>> David Nečas <yeti physics muni cz> ha scritto:
>>>>
>>>>> 1) A few of my objects have some public, documented fields.  They
>>>>> are marked /*<public>*/ and they appear correctly in the docs.
>>>>> But I cannot link to them.
>>>>
>>>> I think the problem is the link to a struct that is subclassing a
>>>> GObject is renamed to MyStructName+"_struct" [1] (resulting in
>>>> MyStructName-struct after the id mangling).
>>>>
>>>> If this is the case, using #GwyField-struct.xreal should work
>>>
>>> You are right, adding -struct fixed it, thanks.
>>
>> As a side note, from the sources it seems not only GObject structs are
>> treated differently, but any struct in the *.hierarchy file (so anything
>> with a _get_type() function present in *.types, I suppose).
> 
> Not everything with _get_type() ends up in .hieararchy.  I'd like my boxed
> types to be listed there as derived from GBoxed -- and their _get_type()
> functions put to <SUBSECTION Standard>, but that does not happen, at
> least for me.

We know that gtk-doc was never properly designed. So please file a bug for this
and then lets figure how to make this consistent without breakign anythiogn and
preferably without another dozen options.

> 
> Also, for some reason FooClass does not end up in -sections.txt
> generated with --rebuild-sections.  I'm not sure if it's my fault or
> a policy to exclude them or a bug (AFAICT when gtk-doc generated
> -sections.txt it has never put FooClass there so it might be a policy).
> I add them with a script at present to be able to document virtual
> methods...

I am not aware of any policy. I think the generated section file should contain
it. Patches welcome.

> 
>> Some time ago I provided a patch to _optionally_ append prefixes to any
>> type unconditionally ("-struct" to any struct, "-union" to any union
>> and so on) but it is still waiting review:
>>
>> http://bugzilla.gnome.org/show_bug.cgi?id=593282
>>
>> This would always permit to have a page with the same name of a type,
>> allowing for example to have:
>>
>> http://library.gnome.org/devel/glib/stable/GString.html
>>
>> instead of:
>>
>> http://library.gnome.org/devel/glib/stable/glib-Strings.html
>>
>> Actually, this is not possible to do.
> 
> Yes, that would be nice.  It's one of the main reasons why I requested
> (and possibly implemented, maybe Stefand did it) @section_id.
> Unfortunately, it can't be used for this purpose due to the id clash.

This is already supported. I just pushed a docs update to mention it. I can't
say for sure who implemented it.

Stefan

> 
> Yeti
> 
> _______________________________________________
> gtk-doc-list mailing list
> gtk-doc-list gnome org
> http://mail.gnome.org/mailman/listinfo/gtk-doc-list