Re: Preserving object props on sockets

Thien-Thi Nguyen <[email protected]> Wed, 04 Sep 2013 22:50:54 +0200
Newsgroups gmane.network.serveez.bugs
Message-ID <[email protected]>
--===============6227170189949499056==
Content-Type: multipart/signed; boundary="=-=-=";
	micalg=pgp-sha1; protocol="application/pgp-signature"

--=-=-=
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

() Thien-Thi Nguyen <[email protected]>
() Wed, 24 Apr 2013 14:40:02 +0200

   See also new branch =E2=80=98q-socket-objprop=E2=80=99, which is just =
=E2=80=98q-upd-userdata=E2=80=99
   squashed and refined a bit (more ready for eventual merge to =E2=80=98p=
=E2=80=99).

   Time to do other stuff now; will revisit in a few days...

After much thought, i've decided that it is best to explicitly NOT
support object properties on UDP and ICMP (generally, connectionless)
sockets.  This is motivated partly by implementation difficulty and
partly by philosophical considerations (i.e., does "object identity"
make sense for such a socket, and if so, could it be possible that
the meaning may be application-dependent?).

I think if the information of interest for the client code is the remote
address (for instance) of a connectionless socket, the client code
should be responsible for extracting that from the socket object and
stick it in a hash table (or whatever), directly.

Does that sound reasonable?

=2D-=20
Thien-Thi Nguyen
   GPG key: 4C807502
   (if you're human and you know it)
      read my lisp: (responsep (questions 'technical)
                               (not (via 'mailing-list)))
                     =3D> nil

--=-=-=
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)

iEYEARECAAYFAlInnTEACgkQZwMiJEyAdQK47wCglzc1Dr8uGN2ManVcPJ/LwzHv
RJcAn2NVxL83DASWWX9MbUxAnUecJYIb
=3t7t
-----END PGP SIGNATURE-----
--=-=-=--


--===============6227170189949499056==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
bug-serveez mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/bug-serveez

--===============6227170189949499056==--