Ahh, the religious war finally arrives on the DC shores
In my experience working with many developers, the K&R style is almost universally derided, while the vast majority of developers prefer
the Allman style.
However, more recently a couple of guys have convinced me of the philosophical superiority of
the Whitesmith style that Mouser referenced. There's no practical difference relative to Allman. But in the context of the BNF for the grammar of C or C++ (see
http://www.nongnu.org/hcb/ ), looking at the definition of
statement and
compound-statement, you can see that what you're indenting is, in any case, some kind of
statement. The Allman approach would have you indenting the opening of a compound-statement at the
same level as its controlling
selection-statement. The Whitesmith style is entirely consistent, though.
From a practical perspective, either of these is superior to K&R because they allow you too comment out things like the controlling selection clause and still get a program that is syntactically correct. But with K&R-style, the opening curlybracket would have been commented out as well, breaking the code.
Let me add this to the controversy: with my style, single-statement blocks are
never legal. Any block, even if it's just a single statement,
must be enclosed in curly brackets. I've seen too many bugs that are the result of adding a second statement to what you'd like to have done in a conditional -- but since there were no curlies, that second statement is executed in all cases. Simply using curlies all the time prevents this from ever happening. (and it's also one reason why python's whitespace-based blocking is better)
And another thing: whenever possible, when doing comparisons, always try to have the lefthand side of the "==" comparison operator be an
rvalue rather than an
lvalue (that is, it's an expression that can't legally be assigned to). This prevents accidental assignment if one of the equals signs is inadvertently omitted.
Recent Posts