Question about new options framework (regression?)

[Baptiste Daroussin]

On Fri, Jul 27, 2012 at 02:25:35PM +0200, olli hauer wrote:
> On 2012-07-27 11:41, Baptiste Daroussin wrote:
> > On Thu, Jul 26, 2012 at 04:41:10PM +0200, Oliver Fromme wrote:
> >>
> >> Jase Thew wrote:
> >>  > On 25/07/2012 23:57, Baptiste Daroussin wrote:
> >>  > > because the priority goes to global to specific and the most specific is the
> >>  > > options file.
> >>  > > 
> >>  > > if most people want the options file to not have the final priority, why not,
> >>  > > can others spread their opinion here?
> >>  > 
> >>  > I can't see why it would be of benefit for saved options to override
> >>  > anything passed to make (either env or as an arg), as one of the reasons
> >>  > you're likely to be passing them is to override any saved settings in
> >>  > the first place.
> >>  > 
> >>  > Please consider reverting back to the established and I daresay,
> >>  > expected behaviour.
> >>
> >> I agree with Jase.
> >>
> >> Actually I'm not sure if PORTS_DBDIR should override make.conf
> >> or vice versa.  I don't know which one should be regarded as
> >> more specific.
> >>
> >> But anything specified on the commandline is definitely more
> >> specific than PORTS_DBDIR and should override anything else.
> >>
> >> One way to do that would be to introduce another pair of
> >> variables, e.g. OVERRIDE_SET and OVERRIDE_UNSET, so you could
> >> type:  make OVERRIDE_SET=STATIC
> >>
> > 
> > I think that is the more reasonnable, I'll add this when fully back. I was
> > thinking of LATE_SET and LATE_UNSET but OVERRIDE_SET and OVERRIDE_UNSET sounds
> > better to me.
> > 
> 
> Why reinvent the wheel ???
> 
> The vars -DWITH(OUT)_FOO is something already well known and documented, the wrapper is already in bsd.options.mk (last entry) but it broken at the moment.
> 

Because WITH(OUT) is inconsistent and is dependant from how the maintainer is
writting its ports: does he check for both WITH_ and !WITHOUT_ for example, does
he check for only one of them?

One of the reason of the new options framework is to get rid of WITH_ and
WITHOUT_ because it is not consistent never work the same over the ports and
that the user have to check the Makefile itself to determine if what is checked
is WITH_ or WITHOUT_

regards,
Bapt
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-ports/attachments/20120727/e9e26fd2/attachment-0001.pgp