including RUN_DEPENDS in TEST_DEPENDS

[Joe Marcus Clarke]

On 12/17/11 7:30 PM, Steve Wills wrote:
> On 12/17/11 18:09, Joe Marcus Clarke wrote:
>>
>> I don't think this is necessarily true for all tests.  Some tests may
>> just validate a checksum or check the build environment for errors.
> 
> I suspect there are cases like that, but I think the vast majority of
> tests that I've encountered do require running the software and having
> RUN_DEPENDS, so I'd rather optimize for the more common case.
> 
>> We're explicit in bringing over those BUILD_DEPENDS that are also
>> required for runtime.  I think we should make porters do the same for
>> TEST_DEPENDS.
> 
> Isn't the reason for that related to packages? That is, a more practical
> concern, rather than simply for the sake of being explicit? I'd rather
> make it easier to use TEST_DEPENDS so that it gets adopted more easily.

Tinderbox should not be helping to drive adoption of a ports feature per
se.  If the intent is for users to explicitly spell out TEST_DEPENDS,
then Tinderbox should enforce that behavior.  My understanding is that
is the case.  If I am wrong, then I would be happy to adopt your patch.

Joe

-- 
PGP Key : http://www.marcuscom.com/pgp.asc