> On Fri, Aug 24, 2012 at 04:32:05PM +0100, Ian Jackson wrote:
> > Some of the information is machine-readable, and some is not. This is
> > obviously necessary in the general case for a description of
> > software's licensing status, since licences are written in human
> > natural languages and might be arbitrary in form.
> Same for debian/control with the Description field.
> We are using it to control several processes anyway. I do not see
> any reason to invent an (at least so-called) machine readable file
> format ( does not say anything from semi-machine-readable) and
> then refraining from using this feature.
AFAICT the machine-readability is being used by a QA post-hoc process
to spot problems, not to control infrastructure.
But on the other hand I think Russ's argument about having all the
information in the same place is very cogent.
So on reflection I withdraw my objection.