GCC Bugzilla Bug 12221

The mentioned line generates compile error:

/u/riesen/gcc-3.4-20030903/lib/gcc/i686-pc-linux-gnu/3.4/../../../../include/c++/3.4/bits/fstream.tcc:
In member function `virtual streamsize std::basic_filebuf<_CharT,
_Traits>::xsputn(const _CharT*, streamsize)':
/u/riesen/gcc-3.4-20030903/lib/gcc/i686-pc-linux-gnu/3.4/../../../../include/c++/3.4/bits/fstream.tcc:463:
error: expected unqualified-id

For such code:

#include <algorithm>
// some header file doing #define:
#define min(a,b) ((a)<(b)?(a): (b))
#include <fstream>

Probably '#undef min' could be useful in every stl header, as vendors and some
long-living projects seem to never catch up with the idea.

Subject: Re:  New: fstream.tcc:463: error: expected unqualified-id

"alexander dot riesen at synopsys dot com" <gcc-bugzilla@gcc.gnu.org> writes:

| For such code:
| 
| #include <algorithm>
| // some header file doing #define:
| #define min(a,b) ((a)<(b)?(a): (b))
| #include <fstream>

This is not a problem in the library.  The code is simply bogus. 
min is a reserved name.  This PR should be resolved invalid.

-- Gaby

Subject: Re:  fstream.tcc:463: error: expected unqualified-id

gdr at integrable-solutions dot net, Tue, Sep 09, 2003 12:18:32 +0200:
> 
> | For such code:
> | 
> | #include <algorithm>
> | // some header file doing #define:
> | #define min(a,b) ((a)<(b)?(a): (b))
> | #include <fstream>
> 
> This is not a problem in the library.  The code is simply bogus. 
> min is a reserved name.  This PR should be resolved invalid.
> 

I'm not saying the library is wrong.

Of course the code _is_ bogus! The problem is, there is a lot of
such code out there.

If the PR is invalid, why there is "undef min" (and max) in the
stl_algobase.h, line 128?

Can there be any help for a sore programmer confronting the problem next
time? Just a hint to help to identify the place where the macro
redefines a reserved word?

-alex

Subject: Re:  fstream.tcc:463: error: expected unqualified-id

"alexander dot riesen at synopsys dot com" <gcc-bugzilla@gcc.gnu.org> writes:

| If the PR is invalid, why there is "undef min" (and max) in the
| stl_algobase.h, line 128?

a carry over from old days, most certainly.  

Probably, should go away.

-- Gaby

Just as Gaby said. You can always place the #undef in your own files, in front 
of the #include. It's not the libraries task to clean up macro names. 
 
W. 

Subject: Re:  fstream.tcc:463: error: expected unqualified-id

bangerth at dealii dot org, Tue, Sep 09, 2003 16:10:58 +0200:
> Just as Gaby said. You can always place the #undef in your own files,
> in front of the #include. It's not the libraries task to clean up
> macro names. 

That's what I did. I'd like to get rid of the offending definition as well,
it's just kind of tricky to find in the maze of headers.

Subject: Re:  fstream.tcc:463: error: expected unqualified-id

"alexander dot riesen at synopsys dot com" <gcc-bugzilla@gcc.gnu.org> writes:

| bangerth at dealii dot org, Tue, Sep 09, 2003 16:10:58 +0200:
| > Just as Gaby said. You can always place the #undef in your own files,
| > in front of the #include. It's not the libraries task to clean up
| > macro names. 
| 
| That's what I did. I'd like to get rid of the offending definition as well,

then that is beyond the scope of V3.

-- Gaby