Re: monikers in a filemanager.

[Michael Meeks <michael helixcode com>]

Hi there,

On 5 Dec 2000, monkeyiq wrote:
> My interest is paeked because I might give in and write my own fm in
> c++. This would not be a competition to nautilus due to using evas and
> fam etc, limiting it to Linux and eye-candy markets. (though I might 
> integrate with nautilus as an option).

        I would encourage you instead to be integrating FAM with
gnome-vfs, and providing a nice CORBA interface via a moniker.
  
> But this could easily be done with a file based paradigm using a
> metafile that contains a moniker... admitadly you require one more
> level of inderection (and one more open/read/close) but my point is
> that using persistent monikers, storing each in its own file, I can't
> see that the vfs method (with file based monikers)  is that much
> weaker than just using monikers for everything.

        Ok; with monikers we represent a massive global object namespace,
whilst I see your point that you would need some sort of base
configuration to get 'root' type anchor monikers ( such as perhaps
) config:/, file:/, smb:\\server\whatever, my_computer:/, desktop:/
etc. storing each sub-moniker in a file would not be sensible, if that is
in fact what you suggest ?

        Also, our moniker implementation is rather different from the MS  
implementation, the only possible serialization of a moniker is as the   
displayName, this saves scads of hassle and brokenness as found in the MS
implementation.

> I have read chunks (most the start of) insode OLE, and many other
> books on COM/DCOM & CORBA. So I should be able to handle
> technical responses.
  
        If you read bonobo/idl/bonobo-moniker.idl you will be able to
contrast our implementation to COMs.
  
        Regards,
  
                Michael.

-- 
 mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot