[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[GNUnet-developers] Re: Block discard question
From: |
Christian Grothoff |
Subject: |
[GNUnet-developers] Re: Block discard question |
Date: |
Thu, 6 Nov 2003 15:14:50 -0500 |
User-agent: |
KMail/1.4.3 |
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Wednesday 05 November 2003 03:30 pm, Igor wrote:
> In manager.c/insertContent(), is it currently
> correct to increment ce->importance by MANAGER_age
> but still make the
>
> if (importance <= dbAPI->getMinimumPriority(computeHighDB(&query))) {
>
> check against nonincremented values on both sides of <= ? It seems
> to me that with the current check there's no actual timeouting. I.e.
> if there is no hidden decrement-all option, the current code
> seems to always prefer a prio 16 content in database to incoming
> content with prio 0, no matter how much time passes?
You are right, it should be importance+MANAGER_age. Fixed in CVS.
> Besides, it throws quite a lot of discards at the moment, I presume
> its pushed content. Could we add a message flag that would inform
> other peers that they don't need to bother with sending
> unrequested content, that is, a disk full flag?
Minus the aging bug, a disk may cease to be full - even for 0-priority
content. Also, it reveils something about your node that you may not want to
expose. Finally, what difference does it make? I mean, instead of padding
with random content, the other peer would now have to send NOISE? How is that
better (minus a minimal amount of processing)? I don't think a disk-full
advertisement would help.
C
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)
iD8DBQE/qqu79tNtMeXQLkIRAm5sAJ9fa98Vz+JHCBtMc9+GLvc/cpEU8wCfUf/x
e3Eg+I7dwHhAwUlFjns4tNE=
=tZC/
-----END PGP SIGNATURE-----
- [GNUnet-developers] Re: Block discard question,
Christian Grothoff <=