Re: schein.debian.org

[Henrique de Moraes Holschuh <hmh@debian.org>]

On Mon, 06 Jun 2011, Bjørn Mork wrote:
> Henrique de Moraes Holschuh <hmh@debian.org> writes:
> > The patch would work, but this is the wrong fix, IMO.  Instead, upstream
> > should change that crap to report the value of "rethops" in the first place,
> > and let the human do the guesswork...
> 
> I agree with you in principle; guessing is wrong...
> 
> BUT, you have to guess this value for the intermediate hops if you want
> the nice "asymm" feature.  And that is IMHO very useful.  I would
> certainly not want to calculate that for each and every hop myself.
> 
> - the guesses are accurate in almost all cases.
> - you cannot guess any better yourself, since you would have to apply
>   exactly the same logic if you were to.
> - the raw TTL values are completely useless information unless you do guess
>   the initial TTL.
> - it's been like this for ages, and anyone making a device sending icmp
>   error messages would know. 
> - if you guess intermediate hops, then you might as well make the final
>   guess, too.
> 
> Even so, maybe it should be more obvious to the common user that the
> application is guessing?  E.g. by adding an option to turn off the
> guessing, or to print raw TTL values in addition to the guessed hop
> count?  Not that I think either would be useful as anything else than
> documentation. 

I'd be fine with printing raw TTL values _in addition_ to the guessed
reverse hop count, plus the change in the patch (to also do to the last
hop what is done for the intermediate hops).

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh