Find dependencies quietly?

[Stephen Kelly]

On 02/02/2014 04:54 PM, Alex Merry wrote:
> On 02/02/14 07:27, Stephen Kelly wrote:
>> On 02/02/2014 12:45 AM, Alex Merry wrote:
>>> The obvious thing to do is set a package property (or something
>>> equivalent) that says it was found as a dependency.  The danger there is
>>> that if something is first found as a dependency and then found
>>> explicitly, it will be supressed.  We could minimise that by also
>>> checking that the TYPE and PURPOSE package properties were not set.
>> What I was getting at is a new TRANSITIVE_DEPEND package TYPE.
>>
>> Existing docs:
>>
>> #     SET_PACKAGE_PROPERTIES(<name> PROPERTIES [ URL <url> ]
>> #                                              [ DESCRIPTION <description> ]
>> #                                              [ TYPE
>> (RUNTIME|OPTIONAL|RECOMMENDED|REQUIRED) ]
>>
>> # If set_package_properties() is called multiple times for the same
>> # package with different TYPEs, the TYPE is only changed to higher TYPEs
>> # ( RUNTIME < OPTIONAL < RECOMMENDED < REQUIRED ),
>>
>> Investigate whether what you want is a separate, similar logic which can
>> 'upgrade' a TRANSITIVE_DEPEND to REQUIRED. Separate because (RECOMMENDED
>> < TRANSITIVE_DEPEND) is not true.
> I would be inclined to say that TRANSITIVE_DEPEND should be below
> RUNTIME, so that any other dependency type overrides it.  

> Either that,
> or have it as a completly orthogonal property that nevertheless
> interacts with TYPE, but I think that's overcomplicating matters.

It is not clear to me whether you realize that that is what I suggested.
You said orthogonal, I said separate. Did you already know that?

> feature(WHAT ALL) should implicitly exclude transitive deps, in the same
> way it implicitly excludes QUIET deps

Yes. Thanks,

Steve.

>  (although I'd be inclined to omit
> the override that QUIET has unless and until someone demands it).