Re: git-diff bug?

From: Junio C Hamano <gitster@pobox.com>
Date: Nov 2, 2020, 22:14 UTC
Message-ID: <xmqqlffjs8ws.fsf@gitster.c.googlers.com>
In-Reply-To: <CALO-gusRt4J5ar45mo7un-EENyt5cX2SQvcXgyMmaHNZg5bFUg@mail.gmail.com>

Eli Barzilay <eli@barzilay.org> writes:

Show 5 quoted lines

>> This matches your results.  That the order makes a difference is a bit
>> odd.  Both are valid diffs of the inputs and neither one changes blank
>> lines, though, so it doesn't look like a bug.
>
> How is it valid?

Just this part. Any patch output that correctly explains how the preimage text changed to the postimage text is a "valid" diff, and that is how René used the word. There are multiple "valid" diff to bring the preimage to the postimage:

    (preimage)          (postimage)
    aaa                 aaa
    bbb                 bbb
    ccc
                        ccc

In an extreme case, this diff is even valid.

    @@ -1,4 +1,4 @@
    -aaa
    -bbb
    -ccc
    -
    +aaa
    +bbb
    +
    +ccc

It's just that it is not as _useful_ as other valid diff that explains how the preimage changed to the postimage.

HTH.