On Tue, Oct 25, 2005 at 09:36:02PM +0200, Spider (D.m.D. Lj.) wrote:
> On Tue, 2005-10-25 at 20:22 +0100, Ciaran McCreesh wrote:
> > On Tue, 25 Oct 2005 20:35:20 +0200 Krzysiek Pawlik
> > <nelchael@gentoo.org> wrote:
> > | Marco Morales wrote:
> > | > I think "none" could be the better workaround imho.
> > | 
> > | I vote for "none" too :) It clearly states, that HOMEPAGE is missing.
> > | Blank HOMEPAGE is misleading.
> > 
> > Then any automated tools will need to be told explicitly that "none" is
> > something special. With all HOMEPAGE entries being fully qualified URIs,
> > you can just do something like for h in HOMEPAGE ; do firefox "$h" & ;
> > done , which works just fine in multiple and blank homepage cases...
> 
> 
> I agree with Ciaran here.  exporting the explicitly empty variable is a
> good thing.  If, however the variable is unset completely, then we have
> an issue of laziness on a developers part.
Why?

portage substitutes "" when metadata keys are unset during depends 
export.  Seems kind of pointless requiring an empty var- yes it could 
be used for typo checks where the dev does a 

HOMEPAGe="blah"

but no other var has an actual check of that sort- to do the check 
would require mangling ebuild.sh also, which I think is kind of daft.

Repoman complains if the var is unset/empty, works for me personally.
~harring