RFS: SDLJump

[Ben Armstrong]

On Fri, 2006-02-17 at 15:36 +0200, Eddy Petrişor wrote:
> The sledgehammer is debtags itself, not the actual tag; no user will
> need without D-I an udeb (at least not one thatknows what he's doing
> ;-), only the installer will use udebs

I have come around to seeing that.  So the rest of my musing has to do
with the inconsistency of having multiple classification systems, not
the particular constraints that appear to be inherent in D-I and udebs.

> the binary udeb package can have different tags than the binary deb,
> with no problems; please document yourself about the purpose of udebs,
> you will understand better.

I did warn you that my question was naive. :)  But yes, I have a general
idea what udebs are for.

> So having an X- header in the udeb will
> not prevent the deb to have an associated debtags tag.

I'm aware that they can differ.  Was my suggestion that such X- headers
be defined in terms of debtags too subtle?  Couldn't the classification
quite easily reside in debtags, and the X- headers be set based on the
debtags tags?

> > I'm assuming a user
> > might care about locating the subset of packages flagged as D-I game
> > material outside of the context of doing an actual install.
> 
> I simply can't imagine that scenario. What would be the purpose?

One springs to mind: communication between users and developers about
the classification.  If the X- header in the udeb is the *sole* place
where this classification is stored, the classification is a bit
obscure.  Only the D-I team itself, plus a handful of people here are
likely to know about it.

On the other hand, if there is a tag, and the X-header in the udeb is
set based on the tag, it is clearer to all people examining Debian's
packages which games are in D-I.  This facilitates decision-making and
teamwork.

Ben