Re: [PATCH] gitlab: Escape git-describe match pattern on Windows hosts

[Philippe Mathieu-Daudé]

On 9/1/21 5:21 PM, Daniel P. Berrangé wrote:
> On Wed, Sep 01, 2021 at 04:17:48PM +0100, Peter Maydell wrote:
>> On Wed, 1 Sept 2021 at 15:59, Daniel P. Berrangé <berrange@redhat.com> wrote:
>>>
>>> On Wed, Sep 01, 2021 at 04:52:29PM +0200, Philippe Mathieu-Daudé wrote:
>>>> Properly escape git-describe 'match' pattern to avoid (MinGW):
>>>>
>>>>   $ if grep -q "EXESUF=.exe" config-host.mak; then make installer;
>>>>     version="$(git describe --match v[0-9]*)";
>>>>     mv -v qemu-setup*.exe qemu-setup-${version}.exe; fi
>>>>   fatal: No names found, cannot describe anything.
>>>>   ERROR: Job failed: exit code 1
>>>>
>>>> Reported-by: Cédric Le Goater <clg@kaod.org>
>>>> Fixes: 8619b5ddb56 ("ci: build & store windows installer")
>>>> Resolves: https://gitlab.com/qemu-project/qemu/-/issues/591
>>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
>>>> ---
>>>>  .gitlab-ci.d/crossbuild-template.yml | 2 +-
>>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/.gitlab-ci.d/crossbuild-template.yml 
>>>> b/.gitlab-ci.d/crossbuild-template.yml
>>>> index 10d22dcf6c1..62d33e6661d 100644
>>>> --- a/.gitlab-ci.d/crossbuild-template.yml
>>>> +++ b/.gitlab-ci.d/crossbuild-template.yml
>>>> @@ -14,7 +14,7 @@
>>>>      - make -j$(expr $(nproc) + 1) all check-build $MAKE_CHECK_ARGS
>>>>      - if grep -q "EXESUF=.exe" config-host.mak;
>>>>        then make installer;
>>>> -      version="$(git describe --match v[0-9]*)";
>>>> +      version="$(git describe --match 'v[0-9]*')";
>>>
>>> Do you have a pointer to a pipeline showing this fix works ?

It worked on my fork but I have some versioned tag:
https://gitlab.com/philmd_rh/qemu/-/jobs/1553450025

Cédric, do you mind testing on your fork?

>>>
>>> It is a bit strange to me. AFAICT, the only difference would
>>> be if the unquoted  v[0-9]*  matched a filename in the
>>> current directory, but that doesn't seem like it is the
>>> case here.
>>
>> We should quote the glob pattern anyway, to avoid possible
>> really confusing behaviour in the future if such a file ever
>> does turn up...
> 
> Sure, I'm happy to see the thing quoted regardless, just want to
> make sure the commit behaviour matches the commit message.
> 
> 
> Regards,
> Daniel
>