comment about gnome architecture

An insightful comment at

The problem is that there are very few people thinking about the
underlying desktop architecture. Someone needs to make a list: what are
the basic services and functions provided by a desktop that will scale
to accomodate any level of integration, function, and coolness that we
can even concieve of. Then impose the architecture at the core of gnome
and build around it. Track this building process as a whole, not as some
far flung module list. Some areas have seen some architecture work, like
gstreamer and icon themes, but for the most part gnome is just a bunch
of disconnected parts all sharing a common name only. This lack of
infrastructure is apparent everywhere in gnome. If someone gets a wild
hair to add some type of feature or structure, then it gets done. But
there is no drive from the top to build a DE framework. Look at as just a simple example of this. All development, news,
documentation is a total mess. Browse if you want a good
scare. There seems to be a shift more and more to build an integrated
whole but still it doesn't seem like anyone is taking that notion to its
extreme. Ideas like the previous posters of having integrated and
centralized ifrastructure that supports contacts, bookmarks,
authentication, plugins, documents, etc just will never be possible
unless the extreme shift to thinking about the gnome architecture is
made. Hacking the ability to access your evolution contacts from gaim is
cool, but it is just a hack compared to the big picture of true
integration and centralization of services. There is no gnome
ifrastructure yet and that is why people have to do crap like that. Why
doesn't nautilus see my galeon or epiphany bookmarks? Because there is
no such thing as "gnome bookmarks". Why are there no "gnome bookmarks"?
Because no one is thinking about integration at that high of a level
yet. This sort of thing is everywhere in gnome and it is no accident. It
reflects the organization of gnome as a whole. 

Why is it so hard to get a gnome mail reader registered by default in a
way that all apps understand? Mozilla cannot integrate with gnome either
by email or app selection because there have never been well defined and
stable interfaces for it to do so. Is it an accident that mozilla on
windows uses 1)the panel? 2)native themes? 3)native print widgets? 4)
default email? 4)default application handlers? No it is not an accident
or thru some preference. Those interfaces are well developed, well
publicized, and stable on windows and mac. Those are the bigges issues
facing gnome, not the lack of a good cd burner. 

The lack of a good gnome RAD IDE is much more relevant than any specific
app. If a well defined desktop infrastructure existed and there were a
really powerful gnome IDE, we would have CD burning apps coming out of
our asses. Same goes for any other app you have been sorely missing in
gnome. If there were a stronger gnome infrastructure the barrier to
producing those apps would be some much lower that we would have them!
The effort needs to be made up front or else it will become harder and
harder to make a change to something cohesive in the future because
there will be so many more users and apps. However, I really have my
doubts that gnome will take off unless these issues are handled. 

The dbus stuff seems like a move in the right direction for that piece
of the gnome infrastructure. It is definitely a "bigger picture" way of
looking at ipc and harware issues. However, how does that fit in with
The big picture? You know THE big picture. There is no THE big picture
so the answer is unknown.

A followup:
I think you don't even really see the BIG BIG picture :) 

The BIG BIG picture is not a GNOME backend for everything, it's a
cross-desktop backend for everything ! 

For example, who wants GNOME bookmarks when you can have cross-desktop
bookmarks usable in GNOME and KDE ? The same goes for a LOT of things.

Marius Andreiana
Galuna - Solutii Linux in Romania

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