Warn on NULL to integer conversion

[chandlerc, 2011-05-26 10:32:04]

http://codereview.appspot.com/4538074/diff/1/include/clang/Basic/DiagnosticSe...
File include/clang/Basic/DiagnosticSemaKinds.td (right):

http://codereview.appspot.com/4538074/diff/1/include/clang/Basic/DiagnosticSe...
include/clang/Basic/DiagnosticSemaKinds.td:1263: def
warn_impcast_null_to_integer : Warning<
to agree with above, 'null_pointer_to_integer'

http://codereview.appspot.com/4538074/diff/1/include/clang/Basic/DiagnosticSe...
include/clang/Basic/DiagnosticSemaKinds.td:1265:
InGroup<DiagGroup<"conversion">>;
Do we want to leave this one off by default until we take a look at how noisy it
is?

http://codereview.appspot.com/4538074/diff/1/lib/Sema/SemaExprCXX.cpp
File lib/Sema/SemaExprCXX.cpp (right):

http://codereview.appspot.com/4538074/diff/1/lib/Sema/SemaExprCXX.cpp#newcode...
lib/Sema/SemaExprCXX.cpp:2056: if (ToType->isIntegerType() &&
!ToType->isBooleanType() &&
This seems a strange location for the warning.... Is there not a place where we
do the rest of the impcast warnings that this one could fit into naturally?

[rtrieu, 2011-05-28 00:20:25]

Switch warning to DefaultIgnore and move checking code so that it is now with
the other implicit cast checks.

http://codereview.appspot.com/4538074/diff/1/include/clang/Basic/DiagnosticSe...
File include/clang/Basic/DiagnosticSemaKinds.td (right):

http://codereview.appspot.com/4538074/diff/1/include/clang/Basic/DiagnosticSe...
include/clang/Basic/DiagnosticSemaKinds.td:1263: def
warn_impcast_null_to_integer : Warning<
On 2011/05/26 10:32:04, chandlerc wrote:
> to agree with above, 'null_pointer_to_integer'

Done.

http://codereview.appspot.com/4538074/diff/1/include/clang/Basic/DiagnosticSe...
include/clang/Basic/DiagnosticSemaKinds.td:1265:
InGroup<DiagGroup<"conversion">>;
On 2011/05/26 10:32:04, chandlerc wrote:
> Do we want to leave this one off by default until we take a look at how noisy
it
> is?

Done.

http://codereview.appspot.com/4538074/diff/1/lib/Sema/SemaExprCXX.cpp
File lib/Sema/SemaExprCXX.cpp (right):

http://codereview.appspot.com/4538074/diff/1/lib/Sema/SemaExprCXX.cpp#newcode...
lib/Sema/SemaExprCXX.cpp:2056: if (ToType->isIntegerType() &&
!ToType->isBooleanType() &&
On 2011/05/26 10:32:04, chandlerc wrote:
> This seems a strange location for the warning.... Is there not a place where
we
> do the rest of the impcast warnings that this one could fit into naturally?

Moved check to the existing check implicit cast function in SemaChecking.cpp

[chandlerc, 2011-05-28 07:10:47]

Looks good with the tweak below... For reference, the rest of the code is
provably false if we have a null pointer constant, so there is no semantic
change to just returning immediately, its just that we needn't do the
computational work to verify the rest.

Go ahead and commit this and mention in the commit log that I reviewed it so no
one wonders.

For reference, I think you can mail these out to cfe-commits the moment you put
them up in rietveld for folks to look at.

http://codereview.appspot.com/4538074/diff/5002/lib/Sema/SemaChecking.cpp
File lib/Sema/SemaChecking.cpp (right):

http://codereview.appspot.com/4538074/diff/5002/lib/Sema/SemaChecking.cpp#new...
lib/Sema/SemaChecking.cpp:2973: << E->getSourceRange() <<
clang::SourceRange(CC);
Do we want to return here?

[rtrieu, 2011-05-29 20:04:47]

This patch has been committed at revision 132297.

http://codereview.appspot.com/4538074/diff/5002/lib/Sema/SemaChecking.cpp
File lib/Sema/SemaChecking.cpp (right):

http://codereview.appspot.com/4538074/diff/5002/lib/Sema/SemaChecking.cpp#new...
lib/Sema/SemaChecking.cpp:2973: << E->getSourceRange() <<
clang::SourceRange(CC);
On 2011/05/28 07:10:47, chandlerc wrote:
> Do we want to return here?

Done.