Re: Building Sketch 0.7.11 under RH 8.0

[Paul Dickson]

On Mon, 13 Jan 2003 12:54:12 -0700, Paul Dickson wrote:

> On Mon, 13 Jan 2003 12:14:15 -0700, Paul Dickson wrote:
> 
> > In file included from _skgtkmodule.c:36:
> > _skgtkmodule.h:30:25: pygtk/pygtk.h: No such file or directory
> 
> I believe this is the cause of the problems.  Compilation in this
> directory stops because of this error.
> 
> For some reason my include paths are set incorrectly.  If I change the
> above include to <pygtk-2.0/pygtk/pygtk.h> then the file is found, but the
> file pygobjects.h can't be found (because it should be referred to as
> <pygtk-2.0/pygobjects.h>).

The rabbit hole is deeper and deeper...

I added the -I/usr/include/pygtk-2.0 to gcc's compilation options.  Now I
think I hit a bug is GCC 3.2.  It won't allow the use of a define within a
structure:

In file included from /usr/include/pygtk-2.0/pygtk/pygtk.h:6,
                 from _skgtkmodule.h:30,
                 from _skgtkmodule.c:36:
/usr/include/pygtk-2.0/pygobject.h:12: parse error before "GObject"
/usr/include/pygtk-2.0/pygobject.h:17: parse error before '}' token
/usr/include/pygtk-2.0/pygobject.h:25: parse error before "GType"
/usr/include/pygtk-2.0/pygobject.h:27: parse error before '}' token

The structure is (lines 10-17):

typedef struct {
    PyObject_HEAD
    GObject *obj;
    gboolean hasref;     /* the GObject owns this reference */
    PyObject *inst_dict; /* the instance dictionary -- must be last */
    PyObject *weakreflist; /* list of weak references */
    GSList *closures;
} PyGObject;

And PyObject_HEAD is defined as:

#define PyObject_HEAD \
        int ob_refcnt; \
        struct _typeobject *ob_type;

I wonder if RedHat used the shipped compiler to compile anything python
related.

        -Paul