[FFmpeg-devel] [PATCHv2] fate: Add an option for disabling the 2k/4k tests

[Martin Storsjö]

On Thu, 12 Dec 2019, James Almer wrote:

> On 12/12/2019 7:03 PM, Martin Storsjö wrote:
>> On Wed, 11 Dec 2019, Martin Storsjö wrote:
>> 
>>> On Wed, 11 Dec 2019, Carl Eugen Hoyos wrote:
>>>
>>>> Am Mi., 11. Dez. 2019 um 09:39 Uhr schrieb Martin Storsjö 
>>> <martin at martin.st>:
>>>>>
>>>>> When testing on a memory limited system, these tests consume a
>>>>> significant amount of memory and can often fail if testing by running
>>>>> multiple processes in parallel.
>>>>> ---
>>>>> Adjusted to use ALLYES instead of a -yes-yes construct.
>>>>>
>>>>> Also moved the 2k tests to the same option.
>>>>> ---
>>>>>  configure             | 3 +++
>>>>>  tests/fate/seek.mak   | 3 ++-
>>>>>  tests/fate/vcodec.mak | 5 +++--
>>>>>  3 files changed, 8 insertions(+), 3 deletions(-)
>>>>>
>>>>> diff --git a/configure b/configure
>>>>> index ca7137f341..922cd8d0ee 100755
>>>>> --- a/configure
>>>>> +++ b/configure
>>>>> @@ -482,6 +482,7 @@ Developer options (useful when working on FFmpeg 
>>> itself):
>>>>>    --ignore-tests=TESTS     comma-separated list (without "fate-"
>>>>> prefix
>>>>>                             in the name) of tests whose result is
>>>>> ignored
>>>>>    --enable-linux-perf      enable Linux Performance Monitor API
>>>>> +  --disable-large-tests    disable tests that use a large amount of 
>>> memory
>>>>
>>>> I would have suggested to control this when running the tests, if the 
>>> configure
>>>> setting makes sense, it should at least be possible to change the
>>>> setting 
>>> when
>>>> calling make.
>>>> Or is that possible anyway?
>>>
>>> It's possible to do e.g. "make fate CONFIG_LARGE_TESTS=no"; any
>>> var=value assignment on the make command line overrides any
>>> var=othervalue assignment within the makefiles themselves, but that
>>> doesn't seem very convenient.
>>>
>>> But I'd like to have it as a configure option, to easily be able to
>>> set it e.g. in a fate setup.
>> 
>> Any further opinions on this one - is it ok to go ahead with it in this
>> form, or are changes requested?
>
> Configure option is fine if it can also be overridden from the command
> line at runtime (like --samples and SAMPLES).

Well, it can be overridden at runtime, but it's not with a very convenient 
name (the CONFIG_* variable). Is that ok?

// Martin