treating OPTIONS

[Thomas-Martin Seck]

* I wrote:

> * Oliver Eikemeier (eikemeier at fillmore-labs.com):
> 
> > Thomas-Martin Seck wrote:
> > >[...]
> > >
> > >Autodetection is not bad as such, but it needs to be overridable and it
> > >should not be allowed to mess with OPTIONS.
> > 
> > I agree that there should be a way to turn it off, but why should the port
> > not preselect OPTIONS that activate features that are available on the
> > current system? I your case you would get LDAP support preselected, but
> > could simply turn it off?
> 
> In my opinion, OPTIONS should be static; it should represent the default
> feature set the maintainer or the software author has/had in mind (that
> is why I do not consider it to be a problem when OPTION's datafile is
> not read in the BATCH and PACKAGE_BUILDING cases because you just have
> to get the parser right, instrumenting the fact that you get a
> WITHOUT_FOO for every WITH_FOO for free). Autotuning this default option
> set can have interesting effects in a package building environment,
> effectively it forces you to build every package in a clean room
> environment to avoid dependency pollution.

Hm, I guess I need more caffeine, sorry: fiddling with presets via
OPTIONS is in fact not as harmful as I originally thought. What we
should avoid at any cost is non-overridable autotuning, especially for
the BATCH/PACKAGE_BUILDING cases.