Re: inconsistency in PROP_TYPE macro declarations ...

[Lars Clausen <lars raeder dk>]

On Tue, 2007-04-24 at 10:29 +0530, Sameer Sahasrabuddhe wrote:
> On 4/24/07, Lars Clausen <lars raeder dk> wrote:
> 
> > > http://svn.gnome.org/viewcvs/dia/trunk/lib/properties.h?r1=3431&r2=3668
> > >
> > > When the property types were changed from PROP_TYPE_REAL to something
> > > else, that broke the PropOffset data structures in some objects such
> > > as objects/standard/line.c
> > >
> > > As a result, properties like line_width and text_height do not get set
> > > in do_set_props_from_offsets().
> >
> > It was done as part of introducing units, so any kind of length or
> > width.  I guess I forgot the ones that are induced by the macros, while
> > the ones that were explicitly defined like data_lwidth in chronoline.c
> > and corner_radius in box.c were fixed correctly.
> 
> OK. I just tried a little grep and perl inline editing ... I think I
> have a patch, I just need to make sure I caught all of them. Should I
> post it here?
> 
> Also, do_set_props_from_offsets() searches for a matching offset for
> each property that it tries to set. It ignores if no offset is found.
> Should we introduce an assertion instead? It might not work if there
> is code that tries to set a property without really checking if it is
> defined. We could a throw a warning in that case, but that could
> result in a lot of clutter on stderr.

Definitely not an assertion, that's way too brutal.  A warning may be a
good idea, and then a round of fixing the current ones.

-Lars