Discussions

using var makes your code less tightly coupled at the source level thanks to homoiconicity, in that respect it's similar to other C# features (extension methods, dynamic, auto-Expression-ization of lambdas). 

for example, maybe you make a bunch of changes to your object model such that the specific types of the intermediate nodes in the Frobnicate -> Tweak -> Confusticate -> * chain change, but the chain is still there, works the same, and the type at the end of the chain stays the same. having used var, this code then won't need to change at all.

Or even the s

8 hours ago, N2Cheval wrote

Or even the sub languages heads like LINQ, RX, Entity, Windows, etc as they have similary problems to deal with in their own way.

in particular somebody should do a panel between the people in charge of all the conceptually similar (in that they're all some variation of "mapping one type system to another", but they all end up feeling like entire frameworks in themselves ...) "data binding/model binding/serialization/schema mapping" parts of WPF, WCF, EF, ASP.NET MVC, ..., with the language people moderating.

I suspect it's a matter of - to mangle Douglas Adams - HTML's fundamental flaws being hidden by its superficial flaws. As long as IE and browser incompatibilities are around for developers to blame their woes on, they'll mask the deeper problems with the web stack.
(Not that WPF/Silverlight don't have their share of flaws as well ...)

The Don Syme session on F# isn't working for me (it's the only one I tried that's not) ... it just gives me a generic error message. It was probably the one I was most interested in too ...

but Win7 was already Vista R2 ...

@USArcher:

25 minutes ago, USArcher wrote

With the exception of games, tell me of one mainstream consumer app that requires or benefits from 2+ cores?  There haven't been any breakthrough applications in the consumer space for years.  No Artificial intelligence, etc. 

personally if there aren't going to be any new breakthrough applications, what's the point of even thinking about this stuff at all?   we should all just quit working in computer science/programming and watch soap operas on TV on the couch all day. actually, even that would be too much effort; if I thought there weren't going to be any new breakthrough applications I wouldn't be able to even get myself to get out of bed in the morning. that's just the way I feel though, you might be right.

1 hour ago, exoteric wrote

It looks like the company has a parallel tracks strategy, a solid evolution of Windows (Windows 7, ...) and this so-far private Midori thing (Longhorn done right?)

I'm a bit worried that it's less a "strategy" and more a rival factions thing. The Midori project still seems to be going strong right now - in fact if anything it seems to be expanding - but eventually it has to come out of incubation and become a product; will it survive the inevitable tension with the existing product teams? Look what happened to Live Mesh ... The current apparent atmosphere of cost-cutting and short-term focus at Microsoft doesn't reassure me.

I think for it to survive they should try and find a narrower concrete use case for it soon. Not replacing Windows, which won't happen for a long, long time, but Microsoft's range of products and services is wide enough that there should be some room for a new kernel designed for distributed systems - maybe powering new services in their datacenters, maybe in their HPC/Technical Computing initiatives, maybe embedded in new devices. Then they can gradually expand to more and more things, and maybe it will become a mainstream flagship OS someday.

41 minutes ago, Bas wrote

That said, I don't give a damn what makes the Windows Phone team happy. I care about what makes me happy

agreed - that's the most sensible policy to have.

Yep, definitely interested in the C#/VB and LINQ sessions and the F# session as well. The biggest frustration I've had with using F# together with C# and .NET projects and libraries is that the various "databinding/mapping/serialization/etc." systems in the various MS .NET frameworks (mapping in LINQ to SQL/EF, model binding in ASP.NET MVC, serialization in WCF, etc etc etc) all seem to assume mutable properties and default constructors. It would be great to either have ORM (RRM?)/services/whatever frameworks designed for F#, or a nice package of adapters etc. Of course I'm having to create some of the latter myself ...

oh and Silverlight for iPhone makes little sense, but Silverlight for Android makes a lot of sense. The time's about right to start talking SL5 too.

For better or worse they are still planning on using Windows as their primary tablet operating system, and putting heavy focus on that for W8. I'm one of the few people who thinks that's probably the right decision, but it's definitely risky: there will probably be different modes for slate and kb/mouse usage, but how much should they share? Make them share too much, and you risk having the interface not work with one or the other or both (the current W7 touch situation), make them share not enough and you risk excessive fragmentation and wasted resources.