Subject: Re: random ip_id must be configurable
To: None <email@example.com>
From: Darren Reed <firstname.lastname@example.org>
Date: 09/16/2003 00:46:28
From Nathan Williams:
> > I don't think most folks are against this being an option that defaults to
> > off. We're against it being on all the time no matter what.
> An option that is off by default won't be used much, and thus won't be
> tested much, and adds to the total amount of complexity. We must
> resist the urge to settle all feature-debates by saying "we can make
> it an option".
Maybe when we get a random id generator that doesn't cause a repeat
nearly quite so early ?
> My impression is that the benefits of this particular feature are
> sufficently small that it's not worth even having as an option.
I'm not sure whether you're trying to say that it should just be
removed or not :), but definately if it is repeating in 38 packets,
that is a serious problem, IMHO.
The more this particular feature is analysed, the more it is clear
that it did not receive sufficient review/testing before it got put
in NetBSD, which itself speaks volumes for the quality process that
OpenBSD uses (or doesn't use.) Maybe core@ should be asked to request
it be removed until it is in satisfactory shape to add ?