>I've heard from some people that the K&R book should be your last read due to the difficulty of it
Speaking as one of those who learned C from K&R, I don't think it's difficult at all. It assumes that you're familiar with certain low level concepts and doesn't offer clarification. A good example is the section on bitwise operators. If you know how to work with bits then it's a wonderful introduction, otherwise you'll be confused about why the operators are needed.
The problem is that too many people want to read a single book and become an expert. That's simply not going to happen. C is difficult primarily due to the brevity of the language itself and the environment in which it evolved (Bell Labs in the 70's). We're talking about a time when everyone knew assembly and the parts of C that confuse you now made perfect sense to the professionals who were picking up the language then. K&R is written for a different audience than those who are looking for an all-in-one fix for basic programming concepts and the C language. That book doesn't exist, by the way. You need several to achieve that effect.
Anyway, whether it's your first book or your last, everyone who claims to know C should own at least one copy. Anyone who claims to be proficient or expert with C should have two or more copies due to wear and tear of the (excellent) binding. I have 5: one early printing for historical reference, two with rubber bands around them to preserve the pages, one on the top of my desk at work and one on the top of my desk at home. The two with rubber rands fell apart because I thumbed through them too often, and my working copies have seen better days despite the care with which I treat my books.
>By the way, if my "language" offends anyone, please tell a moderator to edit it.
If your language gets edited then most of us should fear worse.
>basically the "standard" changes rapidly
True, though the pace of software development means that the "standard" will be used a lot before a new revision or addendum. You can generally expect a change around every five years.
>for example "new" in C++ no longer returns a null pointer on failure but a bad_alloc exception or some crap
Don't confuse standard C++ with pre-standard C++. While ARM C++ was the de facto standard, it wasn't "official" in the way that the ISO standard is. In fact, official standards are typically produced when the threat of diverging dialects begins to really scare people. Besides, you can use nothrow and get a null pointer if it really bothers you.
Personally, I find exceptions to be cleaner to a point.
>yeah, implicit int in main is no longer a standard in c++, not sure about C
Implicit int isn't legal in standard C++ or C99. It's legal, but considered poor style in C89.