On Friday, June 27, 2014 04:25:38 PM Matthew Toseland wrote: > On 27/06/14 16:18, Florent Daigniere wrote: > > On Fri, 2014-06-27 at 16:03 +0100, Matthew Toseland wrote: > >> Re AEAD buckets ... generally authenticated ciphers use a "tag" at the > >> end, equivalent to an HMAC. This means that if there is corruption we > >> will only detect it at the end. In turn this means that *we can't use > >> Closer.close() on authenticated temp buckets!* (We do IIRC) > >> > >> Does this require any action? > > > > Yes, this is bad. We should close() the input stream before > > Closer.close() to ensure we do get any authentication failure related > > exception! > > I think we do in filter code but I'm not sure we do everywhere. Somebody > should look into this.
It might even make sense to someone make Closer not accept such streams... Or just make it not swallow the exceptions from them. I suggest you file a bug immediately about such stuff. Can't do it right now myself, I'm offline. Only able to answer mails due to IMAP... Please do it.
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Devl mailing list Devl@freenetproject.org https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
