Re: [PATCH v3] mergetool: add automerge configuration

[Junio C Hamano <gitster@pobox.com>]

Felipe Contreras <felipe.contreras@gmail.com> writes:

> It doesn't make sense to display easily-solvable conflicts in the
> different views of all mergetools.
>
> Only the chunks that warrant conflict markers should be displayed.
>
> In order to unobtrusively do this, add a new configuration:
> mergetool.autoMerge.

As pointed out by others, I think it makes more sense to have
mergetool.$tool.autoMerge, with optionally mergetool.autoMerge
that can be used as a fallback, and enable it by default.  If
we can make the default of enabling/disabling per tool, that
would be ideal (see my other reply on how to do these).

> +		git merge-file --diff3 -q -p "$LOCAL" "$BASE" "$REMOTE" >"$DIFF3"

Adding "--marker-size 7" to the command line would make the sed
scripts below more robust.

> +		sed -e '/^<<<<<<< /,/^||||||| /d' -e '/^=======\r\?$/,/^>>>>>>> /d' "$DIFF3" >"$BASE"
> +		sed -e '/^||||||| /,/^>>>>>>> /d' -e '/^<<<<<<< /d' "$DIFF3" >"$LOCAL"
> +		sed -e '/^<<<<<<< /,/^=======\r\?$/d' -e '/^>>>>>>> /d' "$DIFF3" >"$REMOTE"
> +		rm -- "$DIFF3"
> +	fi
> +
>  	if test -z "$local_mode" || test -z "$remote_mode"
>  	then
>  		echo "Deleted merge conflict for '$MERGED':"
> diff --git a/t/t7610-mergetool.sh b/t/t7610-mergetool.sh
> index 70afdd06fa..b75c91199b 100755
> --- a/t/t7610-mergetool.sh
> +++ b/t/t7610-mergetool.sh
> @@ -828,4 +828,22 @@ test_expect_success 'mergetool -Oorder-file is honored' '
>  	test_cmp expect actual
>  '
>  
> +test_expect_success 'mergetool automerge' '
> +	test_config mergetool.automerge true &&
> +	test_when_finished "git reset --hard" &&
> +	git checkout -b test${test_count}_b master &&
> +	echo -e "base\n\na" >file1 &&

I think test-lint-shell-syntax should have complained about the use
of "echo -e" here (that's the reason why I queued the patch in 'seen'
initially but it does not appear in what finally got pushed out).

You can either use the plain-vanilla

	cat >file1 <<-\EOF &&
	...
	EOF

because there is nothing gained by saving number of lines with
reduced readability, or can use

	test_write_lines >file1 base "" "" a &&
	...
	test_write_lines >file1 base "" "" c &&
	...
	test_write_lines >file1 local "" "" b &&
	...
	test_write_lines >file1 local "" "" c &&
	...

which would both save number of lines while making it a bit clearer
which 'line' corresponds to which other 'line' in different
preparation of the same "file1".

Will expect a reroll.  Thanks.

> +	git commit -a -m "base" &&
> +	echo -e "base\n\nc" >file1 &&
> +	git commit -a -m "remote update" &&
> +	git checkout -b test${test_count}_a HEAD~ &&
> +	echo -e "local\n\nb" >file1 &&
> +	git commit -a -m "local update" &&
> +	test_must_fail git merge test${test_count}_b &&
> +	yes "" | git mergetool file1 &&
> +	echo -e "local\n\nc" >expect &&
> +	test_cmp expect file1 &&
> +	git commit -m "test resolved with mergetool"
> +'
> +
>  test_done