Re: [avr-chat] Re: WinAvr 2005... compile problems

[Joerg Wunsch]

Chuck Hackett <address@hidden> wrote:

> To me, the big problem is: If you are developing on a Unix flavor
> platform and you actually use the Unix feature that MySource.c is
> distinct from MySource.C you could never compile the product on a
> Windows system because the two distinct files can't coexist.

Nobody [in his right mind] would really do that.  (After all, they'd
all compile into the same object file MySource.o, where the names
would clash.)

However, I've seen a lot of Zip archives floating around in support
forums where the typical Windows developers' sloppyness caused them to
create projects that could *only* be compiled on a case-insensitive
operating system.  (Remember, people post these archives in the hope
to get help on their problem, potentially causing those who are
willing to help additional effort to cleanup the sloppyness.)

To the contrary, I have yet to see a single project created on Unixen
that would fail to also compile on Windows because of filename
problems.

Partially, the problem has been around all the time: the compiler
driver (avr-gcc[.exe]) always makes case-sensitive decisions about
potential filename suffixes.  Of course, it doesn't matter to the
compiler driver that foo.C would reach the same file as foo.c in the
filesystem, yet it would treat the first one as C++ if you spell it
that way on the command-line (and thus, if you spell it that way in
your Makefile).  Likewise, foo.S (spelled that way on the compiler
command-line) will always make the compiler treat it as assembler
source to be preprocessed.

-- 
J"org Wunsch                                           Unix support engineer
address@hidden        http://www.interface-systems.de/~j/