svn commit: r367002 - head/devel/cmake

Adam Weinberger adamw at adamw.org
Tue Sep 2 18:08:38 UTC 2014


On 2 Sep, 2014, at 13:49, John Marino <freebsd.contact at marino.st> wrote:

> On 9/2/2014 19:07, Tijl Coosemans wrote:
>> On Tue, 02 Sep 2014 18:25:15 +0200 John Marino <freebsd.contact at marino.st> wrote:
>>> On 9/2/2014 18:20, Adam Weinberger wrote:
>>>> On 2 Sep, 2014, at 12:11, John Marino <freebsd.contact at marino.st>
>>>>> All this hinges on *if* it is indeed a policy.  If it is, it should
>>>>> be enforced.
>>>> 
>>>> Okay well, until you’re finished writing up your new policy, can we
>>>> make an option to prevent bringing in a dozen unneeded dependencies?
>>> 
>>> That's a pretty unfair statement to make.
>>> I never claimed it's my policy, only that's what I had been told in
>>> #bsdports.
>>> 
>>> On this particular case, these are a dozen "light" dependencies so if
>>> you asking me to vote on it, I say sphinx is not worthy of an exception,
>>> honestly.  The whole set builds in a couple of minutes as I recall.  You
>>> are make it a much bigger deal that it really is, especially given that
>>> these dependencies are most likely already present due to other ports
>>> having required them.
>> 
>> There used to be a NO_INSTALL_MANPAGES variable similar to NOPORTDOCS
>> and NOPORTEXAMPLES.  NOPORTDOCS became the DOCS option and
>> NO_INSTALL_MANPAGES became the MANPAGES option.  Unlike NOPORTDOCS
>> though it was only used when extra dependencies were needed to build
>> manpages, otherwise manpages are always installed.  So in this case
>> it would be okay to add the option.
> 
> 
> So what about all the other ports that use sphinx unconditionally?
> Shouldn't they all be changed to behave exactly the same?  Either they
> all are optional dependencies or all are unconditional dependencies?
> 
> i would have thought that it's not "extra" dependencies, but "heavy"
> dependencies and that it would be an exception to the rule.
> 
> In any case, Adam is right about one thing: This policy needs to be
> documented.

No, you’re right, it’s totally inconsistent. They should all be consistent.

I think you and I have different opinions of what constitutes heavy dependencies. For me, it’s not just the size of them (nobody disagrees about doxygen), but also the number.

I can very easily produce a tarball of prebuilt manpages to accompany the port. Is that an appropriate compromise?

# Adam


-- 
Adam Weinberger
adamw at adamw.org
http://www.adamw.org



More information about the svn-ports-head mailing list