Re: Expanding on NO_ROOT: Categorizing installed files
--ZARJHfwaSJQLOEUz
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Wed, Jul 16, 2014 at 10:46:44AM -0700, Sean Fagan wrote:
> On Jul 16, 2014, at 10:07 AM, Brooks Davis <brooks@freebsd.org> wrote:
> >> I have not changed it from having a new "category=3D" to using "tag=3D=
"; I looked at that, and don't think it's the best way to go, but am still =
looking.
> >=20
> > Given that the current use of tags=3D is basically unconsumed, I still
> > don't understand why.
>=20
> First, because the space is limited -- it's not "tag type=3Dvalue", but "=
tag=3Dvalue". So if there is a category of "debug", that conflicts with a =
tag of "debug." And similarly for any other tags.
>=20
> Second, because "category=3D" isn't the only keyword I might want to add =
here -- I would prefer that the metalog be considered a key-value sequence,=
and any consumer should simply ignore any key it doesn't understand.
>=20
> (Consider checksums, as a semi-obvious example of one that can be put in =
place by install.)
>=20
> Combine the two, and I'm very wary of it -- it puts a limitation in, when=
it should be extensible.
You've convinced me. I think the first is a redherring as no one uses the =
tags, but the second is a good argument.
That said, I think the second argues that -P generating category=3D is the
wrong approach. Instead a new flag should just let you add arbitrary
stuff to the mtree file (subject to validation that it is well formed).
-- Brooks
--ZARJHfwaSJQLOEUz
Content-Type: application/pgp-signature
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iEYEARECAAYFAlPG0VkACgkQXY6L6fI4GtRY/wCfXWU34b2riK+CrP6PxMON1nE0
FpwAoMRnUIX8OZpSbPJrJ2hW3Ljpbbv7
=t4ZM
-----END PGP SIGNATURE-----
--ZARJHfwaSJQLOEUz--
討論串 (同標題文章)
完整討論串 (本文為第 12 之 16 篇)：