Re: Bug 60331 – Object properties should handle multiple objects

[Lars Ræder Clausen <lars raeder dk>]

On Thu, 2008-11-20 at 19:31 +0530, Sameer Sahasrabuddhe wrote:
> On 11/20/08, Lars Ræder Clausen <lars raeder dk> wrote:
> 
> > On Wed, 2008-11-19 at 11:54 +0530, Sameer Sahasrabuddhe wrote:
> >> Hello,
> >>
> >> The current state of this bug is that properties of a group of objects
> >> are handled correctly. The thing left to do is handling properties of
> >> multiple selected objects, that have not been grouped. The problem
> >> arises in app/commands.c, where properties_show() is called ... the
> >> selected objects are stored as a list in DiagramData, but the function
> >> is passed only the head of the list, and all other selected objects
> >> are ignored.
> >>
> >>
> >> The solution is to pass selected objects as a single collection
> >> properties_show(). IIRC, the new version of group will solve this.
> >> What is the status of newgroup.c? Should we wait for this to replace
> >> the current group implementation, or instead come up with a
> >> "DiaContainerObject" whose sole purpose is to forward all function
> >> calls to the objects that it owns?
> >
> > newgroup.c is close to being finished, but I'm not working on it ATM.
> 
> Ok. How do I enable newgroup in the compilation? Didn't notice any
> flag to set ... does it also disable the current group?

Take a look at lib/newgroup.h.  When USE_NEWGROUP is defined, there is
an object in the misc sheet that can be used to explicitly make a
newgroup (parenting). The new group style is really just an extension of
parenting.  This does not disable the old group.

> > So unless someone else picks it up, waiting for it to be finished would
> > be a long wait.  I'm not sure how the DiaContainerObject would differ
> > from the object used in newgroup.c.
> 
> >> "DiaContainerObject" whose sole purpose is to forward all function
> >> calls to the objects that it owns
> 
> That is the only functionality that I need to fix the bug. Will take a
> look at newgroup once I can compile and use it, and see if I can get
> away with enabling only what I need, or else maybe just finish
> newgroup! ;)

Finishing newgroup would be a great thing for Dia.  Frankly, I don't
remember what was missing on it, but I think one issue was in loading
and saving.

> > Notice that newgroup.c fixes
> > another problem with the current group system: It doesn't remove objects
> > from the diagram.  The current group system does, which causes it to
> > break connections leading out of the group.
> 
> Yeah. Basically newgroup helps a lot of things, including the bug
> mentioned here. My dilemma is whether to wait for newgroup to finish,
> or just take what I need.

I would suggest finishing it, if you don't find it too scary.  

-Lars