Re: [PATCH v3 2/3] git-push: clean up some of the output from git push --porcelain

From: Jeff King <peff@peff.net>
Date: Feb 10, 2010, 05:49 UTC
Message-ID: <20100210054957.GG28526@coredump.intra.peff.net>
In-Reply-To: <7vmxzj8hca.fsf@alter.siamese.dyndns.org>

On Mon, Feb 08, 2010 at 02:59:01PM -0800, Junio C Hamano wrote:

Show 13 quoted lines

> >> +	if (flags & TRANSPORT_PUSH_PORCELAIN) {
> >> +		/* Do not give advice messages to Porcelain scripts */
> >> +		advice_push_nonfastforward = 0;
> >> +	}
> >
> > I think this is sane.
> 
> I am tempted to suggest adding "clear_advice(void)" in advice.[ch], so
> that people adding new advices do not have to hunt for even the above
> hunk.  It would be a good direction to go in general _if_ we will have
> more like this --porcelain thing in other parts of the system.
> 
> I didn't do so because that "_if_" is still iffy.

Would it make sense to clear _all_ advice if we are just in porcelain mode for git-push? That is, let's say I am in "push --porcelain" and I try to write a reflog entry for a local tracking ref, but my identity is implicitly picked up from the hostname[1]. Should it trigger advice_implicit_identity and say "by the way, your identity is implicit" on stderr or not?

I would say yes. The advice output should all be on stderr, and the porcelain output should all be on stdout. So there is no parsing conflict. And stderr either goes to /dev/null (because the porcelain is not terminal-based, or doesn't want the terminal screwed up), in which case the advice does nothing, or stderr goes to the terminal (because the porcelain is some simple script), in which case the message is probably something the user would want to see.

-Peff

[1] I am actually not sure if you can trigger the implicit identity advice in this way. But I am arguing from the perspective of "assume you have triggered some random advice through some sub-action", which may include advice to be added in the future.