Deep question for the REAL C GURU's out there ...

[Alma J Wetzker]

Ben Duncan wrote:
> 
> 
> Roger Oberholtzer wrote:
> 
>> On Thu, 2005-02-24 at 15:58, Ben Duncan wrote:
>>
> <SNIP>
> 
>>
>> update the value:    ++(ptr->counter)
>>
>> update the pointer:    (++ptr)->counter
>>
>> get a new hobby:    ++ptr->counter
>>
> 
> Ohh .. that one stings. But then again, precedence of the operators.
> 
> Between you and Kurt and the realization that those using SLAG are
> going to be MORE Applications programmers, I am going to chose
> CLARITY and lots of it, over laziness.
> 
> 
>> It is another one of my quirks. I explicitly specify ordering. Never
>> trust a compiler more than you need to.
>>
>> <SNIP>
>>
> 

FWIW, I have found that most optimizing compilers can do a better job with 
clear code than they can with the other sort.  MORAL, Unless you can write 
better assembler code than a good optimizing compiler, make your code clear.

As far an the coding style issues are concerned, It is best practice to code 
like the rest of the programs in the library you are using.  If you are making 
your own code base, you get to set the style.

My preference for struct's is based on how many places will they be used.  If 
you will make several changes or share them between files, do the typedef 
thing and declare the actual structs in the later code, that way, you can 
easily move the typedefs to an include file.  If you will never use the 
structs outside of the current file, it really doesn't make a difference how 
you declare them, because the compiler is good.

These opinions have an aroma not noticeably different from any other (given 
their source), but I attempted to use logic! ;)

     -- Alma