From: Junio C Hamano <gitster@pobox.com>
Date: Thu, 09 Jun 2011 17:30:18 GMT
Subject: Re: [PATCH] Only print an error for the last connect() failure
Message-ID: <7vhb7yykt1.fsf@alter.siamese.dyndns.org>
In-Reply-To: <767D7D04-6089-4C7B-A532-C5EC9FE0CCC6@apple.com>

Dave Zarzycki <zarzycki@apple.com> writes:

> Personally speaking, I don't think we're masking failures any more than
> git is masking failures when it doesn't find a ref in .git/refs and it
> falls back to .git/packed-refs.

That is not even a fallback. If a loose ref is found, the corresponding
entry in packed-refs is *STALE* and should *NEVER* be looked at.

Back to the topic.

If a host advertises that it can be reached at any of these addresses,
then you can say these addresses are fall-back for each other.  It is fine
not to report about addresses we tried but didn't get connection as long
as we manage to reach it in one of the addresses.  We do not report other
addresses we didn't even try, and it is not useful to report earlier
addresses we tried but didn't get the dial tone in such a case.

As Peff said, if we failed to reach any connection, reporting all
failed addresses would give the user a better clue when diagnosing the
network problem.