>>Do you remember the C++ Journal? The last page was occupied by a column called "Objuscated C++." The objective was to post valid code and let readers try to figure out what it did.
>>
>>I remember Tamar calling C a write-only language.
>>
>
>Actually, it was "obfuscated", but yes, I remember it. There were some really insane examples they had.
>
>However, the "write-only" language is a cliché that was (largely) promoted by those who had written little or no C code.
>
>I'll grant you there was quite a bit of nasty C code out there (and bad VB code, and bad T-SQL code, and yes, even bad Fox code). Certainly the nature of the language (not as high level as others, but not as low level as assembly) and the capabilities probably lended itself more to the kind of crazy examples out there. But the write-only language message was always specious.
>
> And there was also efforts to promote very clean C code. Since you mentioned the C++ journal, you might also remember Dr. Dobbs Journal and the great columnist/author Al Williams - he published a large amount of great C and C++ code in his articles/books. The great developer/author Allen Holub produced an entire C book on how to write a compiler. It was a thick hardcover with a mountain of code - but it was amazingly accessible.

When it comes to BNF ( http://en.wikipedia.org/wiki/Backus%E2%80%93Naur_Form ) C is one of the simpler languages, and consists of relatively few elements. On the other hand, due to the various ways those language elements could be combined (say in the form of declarations), sometimes things did get hairy. Tools like 'cdecl' become rather indispensable. Can't say I'd recommend the method I'd used to learn C -- picked up the K&R book and started plugging away (it did help greatly that I was using a Unix system at the time). Also helps if you're familiar with low-level stuff and have done some form of assembly code before.

Speaking of languages being somewhat tied to a particular environment... It wasn't until I started to use a Control Data mainframe and peculiarities of how data files are structured in the system did I finally "get" why the I/O functions in Pascal worked the way that they did (previous experience was with systems that used ASCII or EBCDIC). And then there was the CDC 6/12 character system -- though in a way it did prepare me for double-byte character systems often used to represent Eastern Asian languages (and if that wasn't bad enough, each language seemed to have their own standard -- a typical example of "standards are good -- everyone ought to have one").