Re: g-ir-scanner not recognising non-virtual functions

On Mon, 26 Sep 2011 15:33:07 +0200
Florian Müllner <fmuellner gnome org> wrote:

> On lun, 2011-09-26 at 13:59 +0100, Tony Houghton wrote:
> > For example, I would like to export this function:
> > 
> > class Gsqlite:Database
> >         from G:Object
> > {
> >     // ...
> From man g-ir-scanner:
> "The suffix determines whether a file be treated as a source file (.c) 
>  or a header file (.h). Currently only C based libraries are supported
>  by the scanner."

gob2 turns that into C (with appropriate .h and .c headers) which is
what I'm feeding to g-ir-scanner. It even includes the annotation
comments. So g-ir-scanner should find this:

 * gsqlite_database_new:
 * @filename: filename of the database or ":memory:".
 * @error: (out) (allow-none) (default NULL):
 * Returns: a new database object or %NULL
 * Use g_object_unref() to close. All statements must be finalized
 * separately.
#line 55 "/media/htpc.home/gsqlite/gsqlite-database.gob"
GsqliteDatabase * 
gsqlite_database_new (const gchar * filename, GError ** error)
#line 178 "gsqlite-database.c"
#define __GOB_FUNCTION__ "Gsqlite:Database::new"
#line 57 "/media/htpc.home/gsqlite/gsqlite-database.gob"
        GsqliteDatabase *self = GET_NEW;
        return self_check_open_for_new(self, self_open(self, filename, error));
#line 187 "gsqlite-database.c"
#undef __GOB_FUNCTION__

All the extras like #line, __GOB_FUNCTION__ [1] and the double braces
are there for at least some of the virtual methods which gir is picking
up, so I don't think they're the cause of the problem.

[1] If I add --warn-all I get this:

gsqlite-database.h:949: Warning: Gsqlite: symbol='__GOB_FUNCTION__': Unknown namespace for symbol '_GOB_FUNCTION__'

which doesn't make sense because __GOB_FUNCTION__ isn't mentioned in the
header and it doesn't have 949 lines; but it seems harmless.

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]