--pgp-sign-Multipart_Wed_Mar_16_10:39:28_2005-1
Content-Type: text/plain; charset=US-ASCII

>>>>> "mh" == Martin Husemann <martin@duskware.de> writes:

    mh> IIUC the problem is that ALTQ did not provide a clean API for
    mh> this,

I think the API is two functions that just translate queue names
to/from tag numbers so that ALTQ and ipfilter can have tag names from
the same namespace.

First there was argument about whether or not it was complicated
enough to be called an API, because if it's not called an API then
it's harder to claim it is significant enough to obstruct commits and
argue about who has to do the work.  Then there was argument about
into which header file to put the prototype.  I don't think the API
itself needs work---it's just a matter of asserting which functions
are in the ``API'' and marking them somehow so they won't regress when
syncing PF/ALTQ to OpenBSD.  see archives around July, 2003.

There will probably be more hard-to-predict work before everyone can
exclude their least-favorite-filter from the system entirely: to make
ALTQ link without PF in the kernel, to make the kernel link without
ipfilter in the kernel, to give 'pfctl' a hardlink 'altqctl' so that
it doesn't have the objectionable text 'pf' in it.  But that above is
the summary of the old argument as I remember it.

--pgp-sign-Multipart_Wed_Mar_16_10:39:28_2005-1
Content-Type: application/pgp-signature
Content-Transfer-Encoding: 7bit

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (NetBSD)

iQCVAwUAQjhTMInCBbTaW/4dAQJfXwQArG2rVab4UKFceNp6qYR7qJhNR4/jIUQ1
NKvLtOAvY+sUZ8OihLaaIj3zhNYRqVQw3JNOc9sqHeqDNSuGujLx9KiKP2FnnF1v
5WXI2lwg/7XrqzaBginwQ8zw2BVcyR6oijb/FxSRYpLICzKDG4ozSD2nN9vT9k+T
kegSzWuXktU=
=abNU
-----END PGP SIGNATURE-----

--pgp-sign-Multipart_Wed_Mar_16_10:39:28_2005-1--