llvm dev - Jul 2011 - [LLVMdev] [PATCH][RFC] upgrade to googletest 1.6.0 (was Re: Build of latest llvm gives warning and error)

On 24 July 2011 10:59, Duncan Sands <baldrick at free.fr>
wrote:
> Hi Edward,
>
>> Building of llvm with gcc-4.5.2-1 gives these warnings:
>>
>> 1) Building FileManagerTest
>>
>> "llvm[4]: Compiling FileManagerTest.cpp for Debug+Asserts build
>>
C:/Programming/VersionControl/llvm/tools/clang/unittests/Basic/FileManagerTest.cpp:
>> In member function 'virtual
>>
void<unnamed>::FileManagerTest_NoVirtualDirectoryExistsBeforeAVirtualFileIsAdded_Test::TestBody()':
>>
C:/Programming/VersionControl/llvm/tools/clang/unittests/Basic/FileManagerTest.cpp:100:3:
>> warning: passing NULL to non-pointer argument 3 of 'static
>> testing::AssertionResult
>> testing::internal::EqHelper<true>::Compare(const char*, const
char*,
>> const T1&, T2*) [with T1 = int, T2 = const
clang::DirectoryEntry]'
>
> yes, they are annoying aren't they.  They all come from google's
unittest
> library.  I think it is best to fix them upstream at google.
We're using googletest 1.5.0. These warnings seem to be fixed in
1.6.0, so here's a patch to upgrade us. Disclaimer: I've only tested
this by doing "make check-all" on Linux/x86_64.

Incidentally, the upgrade would have been much easier if we weren't
carrying local changes to support Haiku OS and Minix. Can these be
pushed upstream (or thrown away if no-one is using them)?

Jay.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: googletest-1.6.0.diff.gz
Type: application/x-gzip
Size: 76389 bytes
Desc: not available
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20110726/c19cd264/attachment.bin>

Hi Jay,
>> yes, they are annoying aren't they.  They all come from
google's unittest
>> library.  I think it is best to fix them upstream at google.
>
> We're using googletest 1.5.0. These warnings seem to be fixed in
> 1.6.0, so here's a patch to upgrade us. Disclaimer: I've only
tested
> this by doing "make check-all" on Linux/x86_64.
>
> Incidentally, the upgrade would have been much easier if we weren't
> carrying local changes to support Haiku OS and Minix. Can these be
> pushed upstream (or thrown away if no-one is using them)?
thanks for doing this.  I vote that we throw Haiku and Minix support away.  If
the Haiku and Minix people want googletest to support these platforms I think
we should insist that support should go upstream first (and we would then pull
the fixes from upstream).

Ciao, Duncan.

On Jul 26, 2011, at 8:57 AM, Duncan Sands wrote:
> Hi Jay,
> 
>>> yes, they are annoying aren't they.  They all come from
google's unittest
>>> library.  I think it is best to fix them upstream at google.
>> 
>> We're using googletest 1.5.0. These warnings seem to be fixed in
>> 1.6.0, so here's a patch to upgrade us. Disclaimer: I've only
tested
>> this by doing "make check-all" on Linux/x86_64.
>> 
>> Incidentally, the upgrade would have been much easier if we weren't
>> carrying local changes to support Haiku OS and Minix. Can these be
>> pushed upstream (or thrown away if no-one is using them)?
> 
> thanks for doing this.  I vote that we throw Haiku and Minix support away. 
If
> the Haiku and Minix people want googletest to support these platforms I
think
> we should insist that support should go upstream first (and we would then
pull
> the fixes from upstream).
+1

-Chris

>>> yes, they are annoying aren't they.  They all come from
google's unittest
>>> library.  I think it is best to fix them upstream at google.
>>
>> We're using googletest 1.5.0. These warnings seem to be fixed in
>> 1.6.0, so here's a patch to upgrade us. Disclaimer: I've only
tested
>> this by doing "make check-all" on Linux/x86_64.
>>
>> Incidentally, the upgrade would have been much easier if we weren't
>> carrying local changes to support Haiku OS and Minix. Can these be
>> pushed upstream (or thrown away if no-one is using them)?
>
> thanks for doing this.  I vote that we throw Haiku and Minix support away.
>  If
> the Haiku and Minix people want googletest to support these platforms I
> think
> we should insist that support should go upstream first (and we would then
> pull
> the fixes from upstream).
I've committed the upgrade to 1.6.0.

I've kept Haiku and Minix support for now (since I'd already done the
merge work), but emailed the relevant parties to try to get them to
push the patches upstream.

Thanks,
Jay.