.Net 3.5 SP1

[Mike]

Now that 3.5 has been released, is there any target for its first SP? My
company, along with many others, is a little leary targeting production
servers with a "1.0" flavor of LINQ and other new features.

Thanks in advance,

Mike

 

[Jon Skeet [C# MVP]]

Now that 3.5 has been released, is there any target for its first SP? My
company, along with many others, is a little leary targeting production
servers with a "1.0" flavor of LINQ and other new features.

I think we (and Microsoft) would need to know what issues there are
before thinking about a target for a service pack...

 

[Mike]

Hi Jon,

Thanks for the reply. I concur, however as in the past, there are always
issues that come up late in the beta cycles that are deemed non-critical to
the final release and resolutions are put off until afterwards and addressed
as a SP. Many, if not most, put off adoption of new technologies (LINQ,
anonymouse types, etc) until its gone through at least one iteration of real
world production release cycles.

Thanks,

Mike

 

[Cowboy \(Gregory A. Beamer\)]

If there is one, it has not been published.

While I agree with you that features are sometimes pushed back, along with
bug fixes, the LINQ and anonymous types features are language features, not
framework features. This is not saying that they will not have any issues,
but that they would not necessarily be tied to a full SP.

Thus far, I am a bit leery about LINQ for reasons other than it not be an
SP1 release. But most of my concerns are Enterprise level.

--
Gregory A. Beamer
MVP, MCP: +I, SE, SD, DBA

*************************************************
| Think outside the box!
|
*************************************************

 

[Mike]

Hi Gregory,

I believe they are tied to both the language and the runtime (only the 3.5
version of csc.exe/vbc.exe can compile them), and the framework was extended
to support them. Just try compiling with 2.0/3.0 as target or under VS2005 -
as you can see, both are required.

Regards,

-Mike

 

[Mike]

An interesting aside to this is just because you target to 2.0, the 3.5
compiler is still used. So if there is a problem with he IL generated by that
compiler, code that worked under VS2005 may not work under VS2008.

 

[Jon Skeet [C# MVP]]

I believe they are tied to both the language and the runtime (only the 3.5
version of csc.exe/vbc.exe can compile them), and the framework was extended
to support them. Just try compiling with 2.0/3.0 as target or under VS2005 -
as you can see, both are required.

You can use almost all of the C# 3 features when using 2.0 or 3.0 as
the target, and when doing so they'll run perfectly well without the
3.5 runtime. Indeed, with appropriate support of 3rd party libraries,
query expressions will work just fine.

The only language feature which absolutely definitely requires 3.5 is
expression trees. Lambda expression to delegate conversion is fine,
anonymous types are fine, extension methods are fine with the addition
of a single extra attribute...

 

[Mike]

Jon,

Correct me if I'm wrong please, but doesnt LINQ make heavy use of the types
defined in System.Linq - such as IQueryable<>, IGrouping<>, etc?

 

[Jon Skeet [C# MVP]]

Correct me if I'm wrong please, but doesnt LINQ make heavy use of the types
defined in System.Linq - such as IQueryable<>, IGrouping<>, etc?

LINQ itself does - but strictly speaking, LINQ isn't a language
feature. The word LINQ only appears in the C# 3 spec in namespaces.

The C# 3 feature *enabling* LINQ is query expressions, and they can
work with anything that provides the appropriate methods. For instance:

public delegate TRet FakeFunc<TIn, TRet>(TIn x);

public class FakeLinq
{
public FakeLinq Where(FakeFunc<FakeLinq,bool> x)
{
return null;
}

public T Select<T>(FakeFunc<FakeLinq,T> y)
{
return default(T);
}
}

class Test
{
static void Main()
{
var x = from z in new FakeLinq()
where z==null
select z.ToString();
}
}

 

[Mike]

Ok, I follow that somewhat. The syntax is simply interpreted as a series of
function calls, etc.

Thanks,

Mike

 

[Jon Skeet [C# MVP]]

Ok, I follow that somewhat. The syntax is simply interpreted as a series of
function calls, etc.

Exactly - and that "query expression to normal C#" translation is
entirely mechanical; it doesn't rely on any specific types etc. That's
how different LINQ providers work - by providing normal or extension
methods which work on their data sources. It's very, very cunning. It
also means that query expressions have a very limited impact on the
language spec - they're an isolated feature, effectively.

 

[Cowboy \(Gregory A. Beamer\)]

While it is true the compilers are package with the framework, I would not
call them a Framework feature, per se, as they are language dependent (csc
versus vbc). I do agree, however, that they ship with a certain version of
the framework, so you will have to have the framework installed to get the
newest bits. It was the language teams who work through features like LINQ,
lambda expressions, etc.

It is true, however, that some features might not compile to 2.0 (if you
target it), but most should not have problems, as the compiler has the
ability to convert most to 2.0. Jon has already covered this.

--
Gregory A. Beamer
MVP, MCP: +I, SE, SD, DBA

*************************************************
| Think outside the box!
|
*************************************************

 

[Cowboy \(Gregory A. Beamer\)]

That is true. I am thinking a bit too Microsoft today.

LINQ syntax, itself, is built into the language. There is much of LINQ that
requires support of the Framework, so I guess that feature does sit on the
cusp of language and Framework.

Back to the original question: I am not sure anyone knows if and when a SP1
is planned and I am sure it would be NDA information if they knew.

--
Gregory A. Beamer
MVP, MCP: +I, SE, SD, DBA

*************************************************
| Think outside the box!
|
*************************************************