the grand conspiracy



Since Michael seems to be convinced that this debate is really about
some hidden agenda, rather than issues of feature sets, code reuse,
compatibility, and so forth, I guess I better explain what he's
talking about, in hopes that we can then return to productive
discussion of the other issues.

Let me quickly summarize what the real story behind 'Hub'

 * Yes, Havoc and I spent a quite a bit of time a few months ago
   researching and writing up a proposal for a shared runtime or
   component system. Distinguishing features being:

    - Very lightweight. Could eventually be used to replace
      systems like GObject and XPCOM.
    - Emphasis on in-process; out-of-process components added
      as an extra layer on top.
    - Able to interact well with existing systems. The vision
      would be that this system would be used across the entire
      range of development being done on open source platforms,
      not just on the desktop.

   The basic scheme looked a lot like UNO, in fact. And is,
   as far as I can tell, quite a lot like what Michaels' 
   vision for where he'd like to end up with Bonobo.

 * We haven't made a big public noise about it because we've been
   busy with GTK+-2.0, and because we _don't_ have code at the
   current time. We did, however, circulate our papers to quite
   a few people for comment, including, as some of the first, Michael
   and Miguel.

 * Unlike Michael, we tend to believe that a radical change to a new
   object system is best achieved by first writing the new object
   system, then moving things over, using bridges for backwards
   compatibility, rather than mutating an existing system by small
   incremental changes that would frequently need to be incompatible.

   Other reasons why we didn't call our system "Bonobo2" were
   a) we believe tying a runtime strongly to any existing system would 
   hinder adoption in areas with vested interests in other existing
   systems b) Bonobo isn't our name to dispose of. 

   So, we used the name 'Hub' to identify the project.

 * Yes, I don't believe that we are to a point where we could
   simply use Bonobo everywhere. Porting GTK+ and Pango to the current 
   Bonobo would be ridiculous --- by its very nature (combining 
   three object systems, one of them being CORBA) - Bonobo is a 
   heavyweight component system.

   But I have not tried to stop anybody from using Bonobo in their
   projects. Neither to my knowledge has Havoc. I believe projects
   like Evolution and Nautilus have gotten significant advantages
   by using Bonobo.

Anyways. for people who believe that bonobo_pbclient_* is a more
standard and natural interface than gconf_client_*, I think it
is wonderful if they have the ability to access the configuration
database through that. 

I hope this description will clear the air enough for people to
realize that there may be other reasons that people like the
GConf system than an aversion to the name "Bonobo".

Regards,
                                        Owen





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]