[cfe-dev] Patch to fix 2>&1 > /dev/null tests on Windows

[John Thompson]

Baptiste,

I had made an error applying the patch, which I had to do manually,
since our versions didn't match up.  I forgot that in Python,
indentation is everything...

I had already resolved the include path issue.  Creating a \dev
directory removed a couple of failures.

Comparing my test failures and yours, it seems you have some failures
I don't have, and I have some you don't have.  Of special interest to
me is that it seems you don't have some of the grep problem I'm
having.  So which grep are you using?  This information would be very
helpful to me.

-John

On Thu, Aug 13, 2009 at 5:23 AM, Baptiste
Lepilleur<baptiste.lepilleur at gmail.com> wrote:
>
>
> 2009/8/12 John Thompson <john.thompson.jtsoftware at gmail.com>
>>
>> Baptiste,
>>
>> Having found I forgot to fix the paths in the count.bat and not.bat files,
>> I tried your patches again, but still had the following results:
>>
>> After add_python_exe_path.patch, no change:  64 failures (no change)
>> After stderr_redirection.patch: 106 failures
>> ...assuming I applied the patch correctly.  I had to do it manually
>> because of differences in our versions.
>> -John
>>
>
> I also needed to create a directory e:\dev to allow /dev/null file to be
> created.
>
> Have a look at your failure, if you see some error such as header "stdio.h"
> or "wchar.h" missing, then you need to change
> InitHeaderSearch::AddDefaultSystemIncludePaths (in
> tools\clang\lib\Frontend\InitHeaderSearch.cpp).
>
> I've pushed my change on https://code.launchpad.net/llvm mirror. Checks the
> win32 branches.
>
> I currently have the following failures:
>
> Failing Tests (45):
>         .\Analysis\casts.c
>         .\Analysis\complex.c
>         .\Analysis\null-deref-ps.c
>         .\CodeGenCXX\new.cpp
>         .\CodeGenCXX\references.cpp
>         .\CodeGen\address-space-field1.c
>         .\CodeGen\address-space-field2.c
>         .\CodeGen\address-space-field3.c
>         .\CodeGen\address-space-field4.c
>         .\CodeGen\cast-to-union.c
>         .\CodeGen\const-init.c
>         .\CodeGen\mandel.c
>         .\CodeGen\mmintrin-test.c
>         .\CodeGen\stack-protector.c
>         .\Driver\dragonfly.c
>         .\Driver\freebsd.c
>         .\Driver\hello.c
>         .\Driver\openbsd.c
>         .\Driver\pth.c
>         .\Frontend\dependency-gen.c
>         .\Lexer\11-27-2007-FloatLiterals.c
>         .\Misc\predefines.c
>         .\PCH\exprs.c
>         .\PCH\reloc.c
>         .\Parser\cxx-template-decl.cpp
>         .\Preprocessor\header_lookup1.c
>         .\Preprocessor\line-directive.c
>         .\Preprocessor\macro_paste_bcpl_comment.c
>         .\SemaCXX\implicit-int.cpp
>         .\SemaCXX\nested-name-spec.cpp
>         .\SemaCXX\nullptr.cpp
>         .\SemaCXX\typedef-redecl.cpp
>         .\SemaTemplate\example-dynarray.cpp
>         .\SemaTemplate\nested-name-spec-template.cpp
>         .\Sema\complex-int.c
>         .\Sema\complex-promotion.c
>         .\Sema\const-eval.c
>         .\Sema\const-ptr-int-ptr-cast.c
>         .\Sema\exprs.c
>         .\Sema\format-strings.c
>         .\Sema\i-c-e.c
>         .\Sema\init.c
>         .\Sema\return.c
>         .\Sema\static-init.c
>         .\Sema\wchar.c
>
> I've checked a few of them and they seems to be real failures. Though, I
> couldn't figure out where to find the current list of failure on buildbot...
>



-- 
John Thompson
John.Thompson.JTSoftware at gmail.com