Screaming Penguin - Source Does Matter

Day Two:

Keynote -- Office 12 / Sharepoint stuff
This is going to be a big thing for microsoft and is geared more towards the IT types. It doesn't really interest me and I had a bit of a hangover, so I zoned quite a bit. Sorry.

It looks like a ton of stuff and I'm sure the little IT/IW weenies came in their pants, but I just tried to stay awake.

Smart Clients
If you're around the MS world at all, you've heard of smart clients. First off, smart clients are fat clients. They are fat clients with "smart" features. Consider it a similar model to the JNLP stuff of the java world, but there are more restrictions/implications. I don't really wanna go into here, but should someone care, I can go into annoying great detail as that is my current job. One of the things that a "smart client" should do is handle disconnected situations. So if you're app uses WS and can't get to the net, then you've got to not crap out, but keep working as best as possible (with cached data, batching updates, handling merges upon reconnection, etc.) This is a royal pain in the ass, and even though MS 'requires' this functionality to be called a 'smart client,' they didn't quite get around to providing a way to do it. (They really can't, it is really a app-domain problem.) They do provide some best practices, though and this is what I sat through. Now that I've typed all this, I can't really come up with something to say other than I got good info for my project.

The next bit that I saw was how to write performant Managed Code. All I really got out of it was all the shit they fixed in 2.0. It was an impressive list, and the best thing is that I don't have to do anything at all. They fixed up som really low-level things like Reflection.

IE overheard...
I got early for one of my sessions and overheard a discussion on IE 7 security improvements so I can expand a little upon how the security works. So, IE runs an untrusted application. This means that anything that is launched by the IE process has this same level of security. This means that no matter what holes IE leaves open, theres nothing it can do to hurt you. As long, of course, as they have the process security crap worked out. So this untrusted process can only touch one directory on your FS and can only launch untrusted processes which would have the same rights. So, you might ask, HTF do you do anything? Well, I'll tell you. What it's got is another broker process that it communicates with. The broker process runs as the current user, so he's got whatever rights you do. Anytime the browser needs to do something with more permissions, it's gotta ask the broker to do it for you. The broker process has acces to HKCU and the user directories on the FS, but not HKLM or other parts of the registry. Anyway, that's the model, we'll see how it works out.

Oh, and atrox, your Macromedia problems are macromedia's fault, not MS. They are putting this shit in places that admin needs rights to. That's not MS's fault.

As a matter of fact, they're pretty much begging people to make sure that the developers make sure that they are coding their stuff to not need admin rights. It takes more work (as it does on any platform,) but they can't explicitly control what an 3rd party dev does.

Vista
Vista is a shit ton more than "fancied up" graphics and searching. Like I said, the searching is by definition more powerful than google's current one because you can add your own metadata that means something to you that, currently, google doesn't allow you to manipulate or search. The UI has been completely redone tossing out win32 and replacing it all together. This means that all the cool shit you can do in other OSes are now possible on windows as well. Also, the underlying communication model of windows is changing from an Event Pump model into a messaging model that is exposed to everyone. Another thing they are doing is (supposedly) eating their own dog-food. They are using all the public APIs to do their widgets. This means that 3rd party component devs get to have all the same power and can use their creativity to what they wanna do. Should be great, really.

The next couple of sessions were of vital importance to me and probably not at all interesting to you, so suffice to say I got hard questions answered with sample code and that makes me sleep better at night.

C# 3.0
Anders himself gave this one (and he's like a frickin rock start to these guys, and admittedly, he seems pretty damned cool.) They're adding 7 features to C# and they are pretty cool. They are adding some really crazy constructs to make the language more flexible and intutive for some "hard shit." I'm not gonna talk about them all cause it'll take me and hour or so. I'll just hit a couple of high points.

Lamba expressions - They are making lamda expressions part of the language. Here is an example of a way to pass an anonymouse delegate to a method that expects a function pointer:

delegate string Foo(Customer c);
public void Bar(Foo);

Bar( delegate (Customer c) { return c.Name } );

stupid example, but there it is. So what they are adding are language statements to consolidate this down to

Bar ( c => return Name )

seems like a syntactic sugar thing, but the real power is that you can assign these guys to variables. So that means you can have code as data (lisp anyone?) I've probably confused more than help, but this is a powerful language addition. you can now treat functionality as data which means that you can code up unknown behavior and allow the plugin data to control the execution. Just cool shit.

Let me add that another really cool thing is that all the language features that they are adding are being targeted towards the existing CLR. This means that the compiler is the only thing that changes. In fact, in the example above, the compiler literally translates the "c => return Name" thingy into the delegate statement. This should help them get RTM a helluva lot sooner.

Lesse, what else? Hell I can't remember now, but when I left that session, I was just stunned. It's really gonna change up how restricted we are in code. I'm sure that Java 1.6/6/3 will have some similar stuff, (hehe, couldn't resist...)

Oh, here's something. The LINQ stuff uses these language features. You can write one LINQ clause and see all the feature additions, (and this is what he did.) As to XPath, they have an XML package, (called XLinq) that can consume XML documents. They are interchangeable with the DLinq( database ORM) and the plain ole object collections. Part of their unit tests include the ability to run (their variant) of all the XQuery W3C examples. What's more, you can write your "Select" functions so that you could really probably override the default selects and implement xpath style queries. It is all very extensible and re-usable. I'm telling you guys, throw away the MS sucks attitude and look at this stuff. It's just plain cool.

After this we went to Universal and had the run of the park. I actually MGM in Orlando quite a bit better.

Anyway, I guess that's it for now, I've got more sessions to get to.

-n