llvm dev - Feb 2006 - [LLVMdev] Compiling with -ansi and -pedantic

Hi,

I'd like to use the compiler flags -ansi -pedantic -Wno-long-long in my 
project but some llvm headers cause errors with them. I tried to compile llvm 
with these flags but it proved to be quite difficult. The headers, however, 
seemed easier to fix. Almost all of the errors were commas at the ends of 
enum lists or extraneous semicolons after namespaces.

I went ahead and hunted down these trivial errors by running each header 
through gcc with -D__STDC_LIMIT_MACROS -ansi -pedantic -Wno-long-long -x c++ 
-S. Obviously a few headers required configuration macros and did not 
therefore compile cleanly, so I might have missed potential errors in those.

Included is a patch against today's CVS that fixes what I managed to find.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: llvm-pedantic-headers.patch
Type: text/x-diff
Size: 19348 bytes
Desc: not available
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20060222/9e1ac577/attachment.patch>

On Wed, 22 Feb 2006, Martin [iso-8859-1] P�rtel wrote:
> I'd like to use the compiler flags -ansi -pedantic -Wno-long-long in my
> project but some llvm headers cause errors with them. I tried to compile
llvm
> with these flags but it proved to be quite difficult. The headers, however,
> seemed easier to fix. Almost all of the errors were commas at the ends of
> enum lists or extraneous semicolons after namespaces.
Looks good, applied:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20060220/032052.html
...
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20060220/032059.html

Thanks for the patch!
> I went ahead and hunted down these trivial errors by running each header
> through gcc with -D__STDC_LIMIT_MACROS -ansi -pedantic -Wno-long-long -x
c++
> -S. Obviously a few headers required configuration macros and did not
> therefore compile cleanly, so I might have missed potential errors in
those.
It would be nice to just fix all of these issues throughout LLVM.  If we 
just built LLVM with those options, you would know that all cases are 
covered, and that no future cases will slip in.

-Chris

-- 
http://nondot.org/sabre/
http://llvm.org/