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

[E. Weddington]

Joerg Wunsch wrote:

> [Please respect the `Mail-Followup-To' header, and do not continue
> discussing this in the avr-gcc-list.]
>  
Technically this probably should have been in the WinAVR-users list. Not 
like it gets a lot of traffic though. But I'll follow up here in avr-chat.


Gary Griswold wrote:

>  
> Let me say that again:  MAKE IS NOW CASE-SENSITIVE!
>  
> Is it to be that way from now on???


David Kelly wrote:

> I hope so, as that is the way things should be. The compiler is case 
> sensitive to value names. It and its supporting tools should be 
> sensitive to file names.


Dave Hansen wrote:

> Sorry, but this is wrong.
>
> "UN*X does it this way" is _not_ equivalent to "This is the way it 
> should be" any more than "Windoze does it this way" is.  
> Case-sensitive file names are an abomination, just as are spaces 
> within file names, and for many of the same reasons.  Just as the fact 
> that "make" treats tab characters differently than spaces, and worse, 
> requires them to function correctly.



Well, there's a sample of opinions.

First off, I'm very sorry that this was sprung on people. My intent was 
to finally upgrade make to the latest version and to have it be a 
completely self-contained Win32 executable, which I was successfull on 
both counts. Unfortunately, I wasn't aware of the case-sensitivity issue 
with make and the underlying file system. Specifically, I wasn't aware 
that there was a configuration switch available, or what the previous 
make was set to.

Now, the main question: Is it to be that way from now on?
Honestly I don't know. Should it be? I have the option to go either way, 
but I'm not completely sure what is the Right Way.

Yes, these tools come from Unix where everything is case-sensitive. 
However they're being run on a file system that is case *preserving* but 
not *case sensitive*. It looks like the configuration switch was added 
by the MinGW folks specifically to handle the stupid Windows file 
systems. So, part of me leans towards that (making it case INsensitive). 
Part of the reasoning is that I agree that it seems silly to have Foo.C 
be required to be a C++ file. I agree in that a filename like that 
should not be a proper C language file and that a programmer needs to be 
more precise. But, people are people, and if there is an easy way to 
avoid confusion, then it make no sense to make it hard either. C++ files 
should be distinct from C files; such as .cxx, or perhaps .cc. That's 
easy enough for me, and I would hazard a guess and say it is easy enough 
for most programmers on the Windows platform too.

So while I agree, in theory, with the Unix folks (and it makes sense on 
Unix where you have to be more precise), Windows is not Unix. And yes, 
as stupid as it is, Windows users (even programmers) don't seem to be as 
precise, don't want to have to be that way, and prefer more hand-holding 
or any other help to  make their life easier.

I would like to hear *thoughtful* opinions about this subject. I would 
prefer that it not degenerate into a religious war of any kind (or I'll 
just simply ignore the thread).

What kind of  GNU Make should WinAVR ship with, and why?

Thanks
Eric