Re: [avr-gcc-list] WinAvr 2005... compile problems

[Dave Hansen]

From: "Graham Davies" <address@hidden>
> Dave Hansen wrote:
>
> > anyone who would create two symbols within a C program
> > that could be used within the same context that differ only
> > by the case of some characters would rightly be ridiculed
>
> Well, I'd better step forward to be ridiculed, then.  I do this all the
> time.  My personal coding style requires non-local variables to begin with
> an uppercase letter and local variables with a lowercase letter.  If I want
> to grab a non-local and do a lot with it then the local that holds it
> temporarily has exactly the property you so deride.  Please ridicule
> constructively if possible.

What a goofball.  ;-)  OK, not too constructive.

But it does sound like a bad idea.  I generally don't make local copies of 
global variables unless it's a volatile value whose value I want to "freeze" 
for some section of code.  But when I do it, I will definitely use a 
completely different symbol rather than one that varies only in the case of 
a letter or few.  You remove the ability of the compiler to diagnose simple 
tpyos or Typographical errors for you.

In my own code, global variables get an all-caps prefix that identifies the 
module from which they come, e.g. TMR_ms might be (is, in several projects) 
a global free-running millisecond counter.  If I wanted to record the value 
of this counter upon entry into a function, I might use something like

  uint8_t timestamp = TMR_ms;

In my code, capitalized symbols tend to indicate macros or typedefs, or at 
least things that aren't what they appear to be, e.g.,

  #define Is_Panel_Moving   BITS_1.bit.b0

where BITS_1 is a global union (defined in the module bits.c) with a uint8_t 
byte field, and a structure of 8 single-bit bit fields named bit.

Regards,
  -=Dave