Re^2: What language should I learn?

[nikosv]

Aside from that C# is a modern and flexible language you might also get to to choose it, or another .NET language for that matter, because of .NET's solid ASP.NET MVC or ADO.NET frameworks

Especially the lack of generic types
generics might be good for most occasions but they cannot do math, so you either have to rely to trickery or,ironicaly, to the dymanic type!

Another criteria for choosing a language is if you can forsee if it will last against time, something that Perl has proven, while you can never be sure with Microsoft.
There is even talk of Dumping .NET and its managed environment as a whole and go back to native code and C++ !

++ for the Roslyn link

[cavac]

The linked article, while quite interesting and certainly at least partly true, does make some weird statements.

To be clear HTML/JavaScript is unproven technology.

Huh... what?!? This August 2011 Web Server Survey found nearly two hundred million active sites running different webservers. That tells me that at least HTTP and HTML are used by millions of people. And this statistics page know of at least 1,659,185 sites using Javascript.

It may be relatively unproven technology for Microsoft, since - for the longest time - they brewed their own soup called "Internet Explorer 6". But in the last few years they gotten rather good at adhering basic standards.

If the future of Windows apps is HTML5/JavaScript then Windows has no future.

What the...? At the time of writing this node, there is a whole caboodle of spezialized hard- and software that only runs on windows which will probably take decades to replace. Microsoft's market share may drop in the regard of "number of windows installation" or they may not. On the other hand, in respect to gaming and office solutions they are a strong contender - especially if they really start pushing HTML/Javascript based SaaS.

As for the notion that client-side operating system choice will get more and more irrelevant: Yes, i hope so. That's why i try to develop everyone of my projects when possible as a web based tool. It saves in continuous porting to different operating systems - and leaves the user a choice, because all (s)he needs is a HTML compatible browser. The application doesn't limit the user preference on the prefered operating system anymore, and i don't have to care either.

Whatever is really going to happen, the next few years will be interesting.

[Jenda]

ADO.Net itself was a big step ... in the wrong direction. LINQ to SQL and Entity Framework/LINQ to Entities may use some part of the ADO.Net libraries internally, but thanks god give you a saner API.

I don't understand what do you mean by your "generics might be good for most occasions but they cannot do math".

The article sounds hugely exaggerated. C# ain't going anywhere any time soon. See eg. Building your first Windows Metro style app using C#, C++, or Visual Basic

See? WinForms are gone, WPF is gone, but the language under the UI may still be C#.

[nikosv]

ADO.Net itself was a big step ... in the wrong direction

I will disagree with that,it is valuable when you want to be in control of your SQL and data access and bindings,while I generally don't like ORM's like EF

I don't understand what do you mean by your "generics might be good for most occasions but they cannot do math"
check Math on Generics with C#’s dynamic keyword

See? WinForms are gone, WPF is gone, but the language under the UI may still be C#.
On the other hand are gone VB6, IronRuby,IronPython,Mono?,Silverlight?,while COM+ is resurected from the dead!History has a habit of repeating itself.And there is a strong push towards C++ as a mainstream language for the win platform

[Jenda]

OK so you'd want something like Clean-style classes. Something like

public T Foo<T>(Whatever<T> param) where T : (+), (-), (*), (/) {
 ...
}

Oh wait ... in Clean this is possible because the +, -, * and / operators are defined as "overloaded", in C# there'd need to be an interface including the four operators and the types that support those operators would have to implement the interface. Then:

// in Clean
foo:: a a a -> a | Arith a
foo x y z = x + y * z

// in C#
public T foo<T>(T x, T y, T z) where T : IArith {
  return x + y * z;
}

A bit more talkative, but as I said ... it's not a problem with the generics per se, but rather with the lack of support for them in the old libraries.

Update: Regarding the ADO.Net as a step in the wrong direction remark: I understand your sentiment regarding the ability to finetune the SQL and have access to the lower levels. The way ADO.Net took though, compared to the old ADO was IMnsHO wrong. While with the old ADO your queries returner recordsets, that could (based on the settings) be taken either as a pipe or navigated freely, with the new ADO.Net you ended up with tons of classes, tables, datasets, datarows, readers, ... It really looked like they expect you to load half of your database to memory and then god know what ... maybe run queries against the local data ... No, I did not like ADO.Net at all.

Now you can treat Entity Framework as just an ORM and let it generate all the SQL based on LINQ queries and hope for the best. Or you can tweak the code generation template and use stored procedures returning exactly what you need exactly how you need, have them propagated all the way to C# in a type safe manner and even have the properties of the returned result rows inherit attributes (even custom ones) you assigned to columns of the tables. Now I do use LINQ and I do let EF generate my SQL ... whereever I'm not yet sure what properties will I need or if the query is simple enough or changes often or I don't care about the speed. If I do care, I define a stored proc, import it into the EF mapping file, tweak the results' attributes if needed and off I go.

There are things I still hate about C#, but it kinda feels I'm going back to the functional languages I used to like at the university. Shame I don't have time to learn F# at the time. Even if I did not end up using it.