[Boost-build] Source character set

Subject: [Boost-build] Source character set
From: Mathias Gaunard (mathias.gaunard_at_[hidden])
Date: 2010-07-24 18:17:20

Data in string and character literals are converted from the source 
character set to the (narrow or wide) execution character set, or to 
UTF-X for UTF-X literals (C++0x).

Therefore, it would be nice if the build system could help in specifying 
the source character set to the compiler. Unfortunately, this is 
compiler-specific, and not always easy.

With GCC, by default, it's the system locale, and this can be changed 
with -finput-charset.
GCC is even nice enough that it allows to change the execution character 
sets as well, but other compilers do not.

With MSVC, by default, it's ANSI (hardly practical).
However, if your source files have a UTF-8 BOM, then the source 
character set becomes UTF-8!
Problem is, if your files have a BOM, GCC will reject them. AFAIK, there 
is no other way to configure MSVC to use UTF-8 as the source character set.

I could imagine it would be nice if Boost.build had a utf-8 option, that 
could automatically prepend a BOM when compiling with MSVC, and send 
-finput-charset=utf-8 for GCC.
I don't know about other compilers, in the worse case it could do 
nothing special.

Does it look like it's a good idea?