Re: There is no .NET in Vista Code?

I.P. Nichols wrote:
"Lurkio" wrote:

I.P. Nichols wrote:

This article is just another example of being able to find "expert opinion" on the web that supports one's bias.

It's pretty much par for the course - BTW, you're not shy
yourself in providing "expert" links in order to support
your own arguments ;-P

I certainly am guilty of linking references in many of my responses but I try (hopefully) not to use contrary opinion sources, after all on this subject I'm clearly in Microsoft's tank. An example of what I'm trying to convey is that if I were railing against say Delphi then I would reference those folks who dislike Borland because they have contrary opinions about how Borland has (or has not) developed Delphi.

Yes, but the points I am making (and those of the article I
linked to) about .NET aren't rants against Microsoft /per se/
(although I /am/ an admitted sceptic <g>) but expressions of
concern at serious flaws we see in the design of the framework.
If those points are fatuous then by all means tear them apart
but complaining about alleged bias and immediately jumping to
negative conclusions regarding the authors credentials makes it
look all the more like they've hit the target and that you don't
have an answer to their questions...

I happen to think that the runtime is a lot better deal than static linking and I'll try to give just one reason and do it without a reference to some "expert opinion".

If you link your program to the .NET dependencies, then forever you are stuck with those modules that were available at the time of linking. I don't know how the folks that develop those .NET linkers think but my bet is they NGen to get rid of the Jitter. Now today with .NET 2.0 there are three different runtimes, one for 32bit and 2 for 64bit. If I develop on a 32 bit machine and static link I foreclose any possibility to have my program know about and take advantage of running on a 64bit system. But if you run it as intended by the framework designers then the jitter will do it's best to optimize for the 64bit CPU in question and the program will make it's calls to a runtime with 64bit routines that take advantage of the 64bit CPU instruction set. That looks like a winner to me.

I definitely take your point about the validity of the
capability to seamlessly "float up" from 32-bit to 64-bit
(though having seen first hand how plenty of un-tethered
..NET apps seem to break when floating up to the latest
framework version, I would want to see this in action
before betting my mortgage on it <g>) but you fail to
acknowledge /my/ central point about the fact that
/there is no single, monolithic entity called .NET/ that
a developer can safely assume will be available on the
target customer machine.

There might be a damn good chance that by this juncture there
will be /some version/ of .NET on the target machine but you
simply cannot make any assumptions that it is the correct
version for your application. Indeed, the lack of guaranteed
backwards compatibility means that you can't even rely on just
specifying that the very latest version of the framework
is installed.

The spiffy new 32bit application you've tied to 1.1 can't
very well seamlessly float up to 64bit when the target
machine has only got 2.0 on it now, can it ? :-P

To answer your question, no I don't think we got here
> through some unforeseen SNAFU. The framework designers
> got us here by thinking ahead with a design that never
> anticipated static linking.

Yeah, well in my opinion they certainly didn't think enough
about the PITA deployment issues it gives to many people out
in the real world...and these are simply going to get worse
as the accumulation of versions increases unless they do
something to address it.


.

Relevant Pages

  • Re: Best .357 Mag revolver
    ... For animals, there will be no idea that mere brandishing the firearm will ... may/may not make for an easier target, ... and you will very likely get a shot from close range on a target ... shouldn't be carrying a firearm for self-defense, in my opinion. ...
    (rec.guns)
  • Re: Clarity vs. code reuse/generality
    ... I'd love your opinion about ... def binary_search(lo, hi, func, target, epsilon): ... return param ...
    (comp.lang.python)
  • Re: Getting back into handguns, need advice...
    ... and target practice-- what is the opinion on these @ ... While the original Sigmas were fairly well panned by many, ...
    (rec.guns)
  • Re: "City Park,CCW Restriction?"
    ... Win a Fulton Armory AR-15 rifle and protect your 2nd Amendment Rights too! ...
    (rec.guns)
  • Re: Evangelicals Role in non sinful Slavery
    ... Ps G and A give the impression of holding that opinion. ... children shouldn't be brought up in any one particular religion. ... to believing that there is a God and going to ... recognise that eliminating all forms of bias is impossible). ...
    (uk.religion.christian)