[
Lists Home |
Date Index |
Thread Index
]
It came time to troll the Longhorn roadmap. Here are things
that I like about it. I know that pages and pages and pages
of blogs and email will be dedicated to the negatives, but here
are positives that stand out to me from a cursory inspection.
1. Customization: some customers have standards that require us to heavily
customize the
look and feel of applications. For MID, we needed a language that easily
enabled us to
control the flow, say navigation, of content and interaction. It was an
IETM language
and for those not familiar with it, that means integration of procedural
testing and
debugging of a tactical asset (say weapon system). So we needed a very
customizable
application interface and we had to deal with the potentials of non-linear
effects
given multiple test assets operating simultaneously (non-linerarity is a
result of
more than one system contending for a resource at the same time). But a
very nice
feature is that I can choose between the browser or a standalone app just by
resetting an attribute and recompiling. It will be fun to work out which is
best
when and where. I think web browsers are a technical cul de sac, but
preserve options.
2. Cheap: That customization has historically required Windows programmers.
Who can afford
that?
3. Partition tasks: Separation of code and the GUI enables us to let
graphics and UI artists design
look and feel and use the objects the programmers develop. We can separate
documents from
applications. Collaborative (say with customers en masse or in pairs )
rapid prototyping
becomes a reality again.
4. Better monitors: Vector graphics scale. Bitmaps don't. When your flat
panel TV is also
your application monitor, you need vectors. The current WinUI must die.
5. Extensible: just try to add your own tags to HTML. See item 1. If an
element is
the surface of a class, it is a lot easier to learn and apply. Suddenly,
XML really
makes sense to an object-oriented programmer. One finally has a reason to
use
a namespace because it tells one what implementation to use as it should.
We get a
realized version of inheritable architectures and that will make some
Hytimers really
happy or really increase their frustration because abstract classes don't
map to tags
and the old Oster dot syntax is back for compound properties.
6. Lifecycle resistant. The old known problem of creating complex
interfaces that
die every time the underlying rendering objects evolve or devolve. This is
coupled
to item 2.
7. State: having an application object to maintain state among pages and
share code
without requiring a server to talk to (a sensible session object). Starting
up
and shutting down an application is easier and more coherent. I can say
exactly
which page starts the food chain. See item 10.
8. Compiled over interpreted code. We did the interpreted style in the US
Navy MID.
It is a non-starter. Just too frikkin' slow. With Longhorn, if I want to
do it
all in procedural code, I can. Policy choice.
9. Improved debug environment. Thank God!
10. Horses for Courses: it knows the difference between a document and an
application.
It knows if an application needs navigation or doesn't and won't insist on
it when it
doesn't. XML isn't SGML On The Web. It's just easier SGML. Navigation
isn't reliant
on URIs. I can navigate to an object, rather than a URI. This should excite
the IETM
vets who worked with test procedure logic and the lost-in-hyperspace
problem. You can
build a smart hub and that is a very neat thing to have (think problems of
degraded
systems that create multiple possible state universes - adaptive topology.
Pre/post
processing is now a slam dunk). Journaling that restores page state is
provided.
11. Neat and very familiar animation operations. Let the games begin. Data
binding
to ANY UI element. Very cool
And on and on. I've run out of time to read the roadmap, but this is very
cool
stuff. Yes, the standardsGurus are going to be very unhappy and so they
should
be. The standardGurus were unhappy with HTML and the emergence of the W3C.
They
are going to be unhappy if those horses are tired or retired. I don't think
that
will happen, but being nervous about changes in our cages is how we know we
are
developers. This should be fun, and for many business reasons, I can
envision
advantages, but it is still three years in the future and that is a long
time
in any business these days.
Still, I'm awfully glad to see Microsoft stick their necks out this far.
Guts
and moxie are thrilling even if risky. Too many vested companies won't take
risks and MS seems to thrive on them.
len
|
