Re: Libole2.

[textshell t-online de]

"Dirk-Jan C. Binnema (djcb)" wrote:
> 
> > The moniker-system would need an additional method to get the uri (or
> > something like that) to the raw data of that object.
> > This extension to the moniker-system is required because the moniker's
> > document structure view maybe totally different from the data
> > organisation and position (e.g. linking).
> 
> You mean a method on monikers to tranfer the type info/raw data of the final chain of the monikers to traverse 'upstream' back to the client at the start. This is quite an extension to monikers... The problems here is  that you break several ( #monikers ) layers of encapsulation.
> 
> So, while using monikers *everywhere* (including in the file manager) has it's benefits, it does require a drastic change to the moniker concept, IMHO.
> 
> > Letting the user choose the aplication to open dthe data with, would be
> > easy with this in place. Just get the uri to the data and it's type,
> > then find a application that can handle that type and pass it the uri of
> > the data.
> > Normaly i think this will only be needed for the last part of the
> > moniker, if all appliction handling that format have the same moniker
> > syntax. Anyway if we can split the monikers at the right places, we
> > could deligate the choice to the user on any level, If a user wishes
> > that degree of choice.
> 
> Needless to say that Nautilus really requires some intimate knowledge of the stuff it's opening, exactly the stuff monikers are supposed to hide...  So, I wonder if at all a fm should use monikers...
> 
> --Dirk-Jan.
> 

I just mean an tight integration of the filemanager/Gnome-VFS with the
object system. As gnome should be a powerful environment i think it
would be nice if the file manager could view the stream and storages of
an Bonobo compound file.
If such a think is implemented monikers would be just a link to the
raw-data for that object as far as the file manager is concerned.  If is
is too complicated to find this storage/stream (eg. the object doesn't
use the standard file format) the VFS could get the storage to operate
on with a bind_to_storage() (refering the recently forwarded moniker
mail) call on the moniker.
I know this would be an really tight integration of the VFS with Bonobo,
but an Desktop-environment's purpose is integration!
So a user can use a moniker just as a filename with gnome. In big
document this could be useful for even replaceing pars as whole
_without_ causing the side-effects that the containing application might
trigger on delete/insert from file.
I know this goes far beyond what is possible in OLE, but ...

 Martin