Re: [PATCHv2 0/13] credential helpers
- From
Junio C Hamano <gitster@pobox.com>
- Date
- Dec 8, 2011, 21:34 UTC
- Message-ID
- <7vmxb2hhne.fsf@alter.siamese.dyndns.org>
- In-Reply-To
- <20111207064231.GA499@sigill.intra.peff.net>
Jeff King <peff@peff.net> writes:
> Because the pattern takes 0 or more lines and no terminator, we can't > distinguish between empty or truncated input and the empty pattern.
I agree that such a positive "Ok here is the end of specification" marker is a good idea, even if we do not worry about "an empty set".
When the requestor wants to specify the credentials with host and user, but the wire is cut after host is communicated but before user is, we do want to notice the communication error, instead of silently erasing all the credentials on the host regardless of the user.
Thanks.