From: Junio C Hamano <gitster@pobox.com>
Date: Tue, 23 Sep 2014 20:17:18 GMT
Subject: Re: [PATCH for-maint] apply: gitdiff_verify_name(): accept "/dev/null\r"
Message-ID: <xmqqoau6hz1t.fsf@gitster.dls.corp.google.com>
In-Reply-To: <xmqq1tr2jhg2.fsf@gitster.dls.corp.google.com>

Junio C Hamano <gitster@pobox.com> writes:

> SMTP transport may be CRLF-unsafe, so I have a suspicion that it may
> turn out that what you are trying to do might be an equilvalent of
>
> 	git format-patch ... |
>         # first lose all \r\n
>         dos2unix | 
> 	# then make everything \r\n
>         unix2dos |
>         # and apply
>         git am
>
> which is not workable in the first place.  I dunno.

This is a tangent, but if the problem were slightly different, I
would be more sympathetic.  For example

    A popular MUA, when asked to write out a message in the mbox
    format, ends _all_ the lines in its output with CRLF, whether
    the original was sent as LF-only or CRLF, and there is no way to
    convince it to use LF-only.  "git apply" fails to grok such an
    input.

could be, if the use of such an MUA is very prevalent, a common
problem worth working around.

But then I would suspect that the workaround for such a case may not
be "accept /dev/null\n and /dev/null\r\n equally".  It is likely
that the right workaround for such a case would be to "turn all \r\n
into \n before processing".