[xmlroff] Simple docbook

[Tony Graham]

On Tue, Mar 18 2008 21:43:56 +0000, asmodai at in-nomine.org wrote:
> -On [20080318 22:05], Tony Graham (Tony.Graham at MenteithConsulting.com) wrote:
>>I still haven't found where the innocuous looking G_PARAM_CONSTRUCT_ONLY
>>was causing the problem, but it was.  AFAICT, every g_object_new() for a
>>property also set the property's value.
>
> Cool Tony,

I'm glad you think so, since I was worried that it was impenetrable
gobbledygook.

For those to which it was, the problem was that the
G_PARAM_CONSTRUCT_ONLY was used to set the "value" (GObject) property of
the (GObject) FoProperty so that "value" could only be set when the
FoProperty is created.  FoProperty is the parent class of all XSL
property classes, so this affected all properties, hence the succession
of NULL assertion failures.

(GObject) properties can be set when the GObject is created with
g_object_new(), and properties can be set after object creation with
g_object_set().  There's many instances in the libfo code of
g_object_new() for properties where the "value" (GObject) property is
also set, and there's no instances of g_object_set(), so I'm stumped as
to why it was a problem.

I'm also surprised that there were no error messages about "value" being
set after construct time, if that was the problem.

> I'll give 'r a spin tomorrow when at work.

Thanks.  I'll have limited Internet access between now and Saturday, so
I may be slower to respond until then.  (FWIW, I'll be in Bristol, UK.)

Regards,


Tony.