[PATCH] D54942: [PDB] Make PDB lit tests use the new builder

[Zachary Turner via llvm-commits]

I was thinking that we could just automatically compute the output file
names as:

os.path.join(out_dir, basename(output_file) + '.' + basename(input_file) +
'.obj')

Currently it's just

os.path.join(out_dir, basename(input_file) + '.obj')

which is why I think the problem occurs.

On Wed, Dec 5, 2018 at 12:47 PM Aleksandr Urakov <
aleksandr.urakov at jetbrains.com> wrote:

> With such solution there would be even no need to change the current
> commit. But I'm not sure that it's trivial to do - the output file name may
> contain path with directories. May be we can replace slashes with
> underscores in the output file path and concatenate it with the object file
> name? Or even replace slashes in the source file path and concatenate it
> with the output file path - so object files will be located in the same
> place as the output file.
>
> Am Mi., 5. Dez. 2018, 23:30 hat Zachary Turner <zturner at google.com>
> geschrieben:
>
>> It is not possible to specify object file name in compile and link mode.
>> But perhaps we can just change the default object file name to include
>> something from the output file as well
>> On Wed, Dec 5, 2018 at 12:26 PM Aleksandr Urakov via Phabricator <
>> reviews at reviews.llvm.org> wrote:
>>
>>> aleksandr.urakov added a subscriber: zturner.
>>> aleksandr.urakov added a comment.
>>>
>>> The similar problem with `typedefs.test` is here:
>>> http://lab.llvm.org:8014/builders/lldb-x64-windows-ninja/builds/1940/steps/test/logs/stdio
>>>
>>> I have an assumption about the cause of the problem. Are the tests
>>> running in parallel? In this case `typedefs.test` and `enums-layout.test`
>>> are writing to the same object file together, because they both are
>>> compiled from the same source.
>>>
>>> @zturner Is it possible to specify object file's name in
>>> `compile-and-link` mode? Then we can specify different names in different
>>> tests for both object files and executables. But I think that splitting the
>>> source or combining the tests would be a better idea.
>>>
>>> I will fix it only tomorrow, because I'm already at home today. Feel
>>> free to revert it if needed.
>>>
>>>
>>> Repository:
>>>   rL LLVM
>>>
>>> CHANGES SINCE LAST ACTION
>>>   https://reviews.llvm.org/D54942/new/
>>>
>>> https://reviews.llvm.org/D54942
>>>
>>>
>>>
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20181205/d9dba917/attachment.html>