I find it ironic that you dislike C++11 but the example you link to is only bloated due to the lack of the very same features that C++11 adds (in this case variadic templates would make that file 10 times smaller).
And it only really exists in the first place to paper over the lack of lambda functions, which C++11 also added.
Modern C++11 code is much easier to read and understand. The majority of added features are all about promoting that very ideal.
http://qt.gitorious.org/qt/qtbase/blobs/59fb0328392a5fc47e3a...
(search for the #else bits)
Some of the boost libraries actually demonstrate this very well. If you look at the implementation of the library, you may be intimidated by its complexity (as you have shown with bind), but if you look at examples of its usage, you will often find it a lot cleaner than other languages.
Imagine what you would have to do in Java in order to do what bind does:
int diff(int a, int b) { return a - b; }
auto reverse_diff = bind(diff, _2, _1);
assert(diff(5, 2) == 3);
assert(reverse_diff(5, 2) == -3);
(Obviously, this particular example is trivial in any dynamic language, but that's a different debate)I'm not saying C++ is the only or even the best language for creating statically typed DSLs. But it is a language in which you can optimize for readability in a much more radical fashion than in some other mainstream languages.
A lot of the code that tries to showcase new C++ features often falls short on some of the intricacies of the language or doesn't embed it into a larger scheme of a library. If my library offers customization via ADL across the board that kind of coding goes right out the window.
I'm aware of the proposals to make function objects through ADL, but this is but one example where the complexity of C++ comes to the surface.
That said, I don't like the fact that each and every feature of C++ comes with a really long list of weird special cases that are impossible to keep in your head.
And you WILL have to debug it. Even if your code is perfect, one day someone will stomp on your memory and you'll be looking at this stuff in gdb, cursing the day you ever heard of Boost.
In Java, however, lots of things are done outside of the language to solve similar meta-programming issues. For instance, many of the new annotations are only available within a suitable container that interprets them at runtime. Some of these things even need byte code manipulation. The resulting system is not generally simpler than C++ in my opinion and you have the same separation between library/container/runtime makers and mere users.
Why? It's so damn easy for me to understand.
> I also dislike perl. A large amount of perl code contains
> shorcuts - like $^p or %#p (not real perl patterns but I
> hope you get the point).
Well, that's not so fair - as that's... not (capital "P") Perl. That's like mimicking how a foreign language sounds like to you, (without context)http://www.youtube.com/watch?v=Q1VGoKBKR3I
and declaring that language unreasonable. So no, I don't understand :)
I owe quite a bit to Perl. I couldn't, ever, learn C++.
Like I said, writing code in any language is easy. Making it readable for others is the hard part.
Sadly, that was also the day I gave up on learning Perl, since I am the maths kind of guy and I get hopelessly lost if my programming language isn't built on solid logical ground, with few "core" features that I can actually aim to master.
Assuming this is a real problem for you, and you aren't just looking for arguments to validate your dislike: `perldoc perlvar` and then search for the variable.
When I was doing perl, I used a lot of special variables. I don't see why you shouldn't know what $_, $!, $0, $? do when you are programming perl. I avoided the lesser knowns, but that is how it is in every language - there always are obscure features and it's the programmers responsibility to write reasonable code.
I am a good C++ developer but my dislike for the language has grown ever since c++11. Many parts of the language are arcane and unfortunately these parts are used over and over again by many programmers either to show of their 'coolness' or because it's the only possible way.
Here try this for size: https://github.com/boost-lib/bind/blob/master/include/boost/... (sorry if that blurb made you blind)
I also dislike perl. A large amount of perl code contains shorcuts - like $^p or %#p (not real perl patterns but I hope you get the point). These patterns are ungooglable and hard to remember. The only people who know them are those who have spent a lot amount of time with the language.
The result is code that is extremely hard to understand and read.
Languages must be optimized for reading and not writing. Of course, nobody is asking one to go overboard (like COBOL) but reading a computer program should not hurt the brain and should not require years of training.
This is why I will not advocate C++. I would advocate C, Python or JavaScript (yeah javascript, it has it's pitfalls but it's very easy to master). There are possibly other languages like Ruby but I haven't taken the time to learn them.
(BTW, If you want to see how C++ could be done right, look at something Qt from an API point of view).