From: Junio C Hamano <gitster@pobox.com>
Date: Thu, 30 Jun 2022 21:03:22 GMT
Subject: Re: [PATCH] git-send-email: Add --no-validate-email option
Message-ID: <xmqqzghtdftx.fsf@gitster.g>
In-Reply-To: <220630.868rpee6d3.gmgdl@evledraar.gmail.com>

Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:

> I'm suggesting that we replace our own validation with that of the SMTP
> server's, yes they're don't 1=1 correspond, but I think the part of the
> Venn diagram of where that matters is too small to worry about.
>
> It has the advantage of side-stepping issues with not having
> Email::Valid, as well as those cases where we're being overzelous about
> RFC validation, but our local SMTP is willing to try to deliver the
> mail.
>
> It's not like authors of MTAs haven't heard of that character limit, but
> they're also aware that that certain parts of the spec are loosely
> enforced, and that trying delivery is often better than rejecting a mail
> out of RFC pedantry.

I am not sure if that is a healthy direction to go.

If a local outbound relay is written with the knowledge that it will
never be talking to the SMTP at the final mailbox directly, I would
expect that it may not implement any validation at all, relying on
the "next hop" smarthost to reject anything invalid it throws at it.

So...