svn commit: r321445 - head/mail/gnarwl

[Adam Weinberger]

On 2013/06/21 03:38, Alexey Dokuchaev wrote:
> On Thu, Jun 20, 2013 at 11:23:05PM -0700, Jason Helfman wrote:
>> > -CONFIGURE_ARGS+=--with-docdir=${DOCSDIR}
>> > +CONFIGURE_ARGS=        --with-docdir=${DOCSDIR}
>> 
>> It may look more pretty, but in the end your original commit didn't 
>> clobber
>> any pre-existing CONFIGURE_ARGS that may have been set by the user. 
>> This
>> one does.
> 
> We never supported preset (users) CONFIGURE_ARGS, and it would be silly
> thing if we did (contrary to CC/CFLAGS).  There is an umpteen number of
> configure scripts out there, both coming from different, often 
> incompatible
> versions of autotools or written manually, and handling any 
> CONFIGURE_ARGS
> in /etc/make.conf would likely break a lot of ports.  User is not 
> supposed
> to mess with them on that global level.
> 
> It's maintainer's duty to ensure that port provides appropriate 
> configure
> arguments; if user is smart/brave enough to change them, we must assume
> that we knows what he's doing, and can edit Makefile directly.  That 
> said,
> CONFIGURE_ARGS= is correct here.

As a smart and brave user, I prefer to be able to fiddle with things in 
/etc/make.conf,
so that I don't have to re-patch the Makefile every time it gets 
updated. And sometimes,
I even want to make customizations that aren't provided as an OPTION. A 
nice way to do
that is
.if ${.CURDIR:M*/mail/someport}
CONFIGURE_ARGS+=    --with-custom-banner=something
.endif

I understand your point, but if we're going to have a perspective, why 
not make it the
perspective that lets users do as MUCH as possible, rather than limiting 
them to the
subset of things that we pre-think for them? Our users are mighty smart 
folk; I propose
giving them the benefit of the doubt.

# Adam


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