[krbdev.mit.edu #3685] EOF mistakenly interpreted as error causes re-use of closed stream

[Russ Allbery via RT]

rainer weikusat at sncag com via RT <rt-comment at krbdev.mit.edu> writes:
> Russ Allbery <rra at stanford.edu> writes:

>> Yup, this is RT #3549, which I think includes a slightly better patch
>> courtesy of Steve Langasek.

> Presumably, because it returns a totally random synthetic system error
> that bears absolutely no relation to the condition that caused it?
> (keytab w/o kvno)

Well, no, more because I think it's cleaner to detect EOF from the
function return than explicitly call feof.  It may well be that your
choice of error code is better; that's a very good point.

> Do me a favor and *DO NOT* send me any of this "Hey, I know this guy but
> who an earth are you"-nonsense again, ok?

It would be useful to be able to have a conversation about the merits of
an approach without this sort of reaction.  I was not intending to be
offensive; I was intending to point out for the Kerberos developers that
this should probably be merged with the other bug report for the same
issue and was advocating a patch that I thought was cleaner.

If you disagree and have good reason, as it sounds like you do, then by
all means we should arrive at the best possible patch!

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>