Format inserted/deleted maps in list assertions #13954
Conversation
|
@sabiwara You're right, I was actually testing it with this code because the output of the test function was hard to read . When I swap the left and right it goes a different code path. Will investigate further 484 assert [map, %{map | hour: 8}] == [map] #this one is called
485 assert [map] == [map, %{map | hour: 8}] #this one is not |
| @@ -469,6 +469,58 @@ defmodule ExUnit.FormatterTest do | |||
| """ | |||
| end | |||
|
|
|||
| test "formats nested maps with column limit" do | |||
There was a problem hiding this comment.
Could we have a test in diff_test.exs instead?
lib/ex_unit/lib/ex_unit/diff.ex
Outdated
| end | ||
|
|
||
| defp move_right({y, list1, [], edits}) do | ||
| {y, list1, [], edits} | ||
| end | ||
|
|
||
| defp move_down({y, [elem1 | rest1], list2, {edit1, edit2, env}}) do | ||
| {y + 1, rest1, list2, {[{:del, elem1} | edit1], edit2, env}} | ||
| {y + 1, rest1, list2, {[{:del, build_elem(elem1)} | edit1], edit2, env}} |
There was a problem hiding this comment.
I am worried this change is happening too deep in the algorithm. We should probably deal with converting to struct/map when building the diff script?
There was a problem hiding this comment.
I can try to move it where it is extracted from the diff later, it still may be not the best place.
There was a problem hiding this comment.
I'ts enough to convert it in the escape function, it is called in the diff_improper when the non matched elem is inserted on the right (line 401) and list_script_to_diff when it's in :del on left side on line 621
|
I had to restrict the width from infinity to 80 chars to reproduce the error. |
| @@ -126,10 +126,11 @@ defmodule ExUnit.DiffTest do | |||
| assert env_binding == expected_binding | |||
| end | |||
|
|
|||
| @terminal_width 80 | |||
There was a problem hiding this comment.
this changes other tests. I can either adjust it or make a custom helper to test it ???
Solves #13624 When the elements are equal then we pass them as ast, when deleted or inserted we pass maps as
%{a: 1, b: 2}and fall into inspect in safe_to_algebra. This pr changes maps to ast representation so it's formatted correctly in the function that handles maps.I wrote a test, but I couldn't make it fail initially. However, it did fail when I added a regular assertion in my code. There are likely similar issues with other data structure, but now that I know what to change, it should be easier to extend the fix. It wasn't easy to find the issue with all the pattern matching in function heads and the single-element tuple which is not that easy to spot. It would be helpful to have a
dbgimplementation for multiple function heads.