[PATCH] Fix test-org-element/cache

[PATCH] Fix test-org-element/cache

[Ihor Radchenko]

[-- Attachment #1: Type: text/plain, Size: 1285 bytes --]

Hi,

I encountered a bug in test-org-element/cache. The test creates the
following Org file with table and runs (org-element-at-point) at the
beginning of the second line. The expectation in the test is that we are
at the table row, while in reality these are two distinct tables (the
"ab" line is a table within "item" list entry):

#+begin_center
P0

- item

  P1
  | a | b |
| c | d |
#+end_center

I tested with emacs -Q (with no org-element-cache enabled). The return
value of (org-element-at-point) is (table (...)). In the test (before
inserting "item") the same point has different element:

#+begin_center
P0



  P1
  | a | b |
| c | d |
#+end_center

Now, org-element-at-point (correctly) returns (table-row (...)).

The master version of the test assumed that element would not change
after modification (wrong). Yet, the test was passing with
org-element-use-cache enabled (it should have not been).

The current patch is fixing the test. Though I am not sure if it should
be applied right now. The correct test will fail and org-element also
need to be fixed. I do have a fix for org-element, but it is quite
complex, not fully ready yet, involves other features, and needs to be
discussed carefully. So, I am only submitting the patch for the test for
now.

Best,
Ihor


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Fix-test-org-element-cache-Preserve-local-structures.patch --]
[-- Type: text/x-diff, Size: 1403 bytes --]

From b9da1e2c4faeb5545fb43641209dd6eca80f5d51 Mon Sep 17 00:00:00 2001
Message-Id: <b9da1e2c4faeb5545fb43641209dd6eca80f5d51.1621077736.git.yantar92@gmail.com>
From: Ihor Radchenko <yantar92@gmail.com>
Date: Sat, 15 May 2021 19:19:09 +0800
Subject: [PATCH] Fix test-org-element/cache "Preserve local structures when
 re-parenting."

* testing/lisp/test-org-element.el (test-org-element/cache): Add
spaces to make the "abcd" table a single table.  The old version uses
example org text with two tables and was passing because of another
bug in org-element-cache.  Testing the old version with
org-element-use-cache set to nil would yield error, while it should
not.
---
 testing/lisp/test-org-element.el | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/testing/lisp/test-org-element.el b/testing/lisp/test-org-element.el
index 4f108754c..28e9ecd12 100644
--- a/testing/lisp/test-org-element.el
+++ b/testing/lisp/test-org-element.el
@@ -3770,7 +3770,7 @@ (ert-deftest test-org-element/cache ()
    (eq 'table
        (let ((org-element-use-cache t))
 	 (org-test-with-temp-text
-	     "#+begin_center\nP0\n\n<point>\n\n  P1\n  | a | b |\n| c | d |\n#+end_center"
+	     "#+begin_center\nP0\n\n<point>\n\n  P1\n  | a | b |\n  | c | d |\n#+end_center"
 	   (save-excursion (search-forward "| c |") (org-element-at-point))
 	   (insert "- item")
 	   (search-forward "| c |")
-- 
2.26.3

Re: [PATCH] Fix test-org-element/cache

[Bastien]

Ihor Radchenko <yantar92@gmail.com> writes:

> I encountered a bug in test-org-element/cache. 

Applied in maint, thanks!

-- 
 Bastien

Re: [PATCH] Fix test-org-element/cache

[Nicolas Goaziou]

Hello,

Ihor Radchenko <yantar92@gmail.com> writes:

> I encountered a bug in test-org-element/cache. The test creates the
> following Org file with table and runs (org-element-at-point) at the
> beginning of the second line. The expectation in the test is that we are
> at the table row, while in reality these are two distinct tables (the
> "ab" line is a table within "item" list entry):
>
> #+begin_center
> P0
>
> - item
>
>   P1
>   | a | b |
> | c | d |
> #+end_center
>
>
> I tested with emacs -Q (with no org-element-cache enabled). The return
> value of (org-element-at-point) is (table (...)). In the test (before
> inserting "item") the same point has different element:
>
> #+begin_center
> P0
>
>
>
>   P1
>   | a | b |
> | c | d |
> #+end_center
>
> Now, org-element-at-point (correctly) returns (table-row (...)).

I'm not sure to understand your report, but `org-element-at-point'
should return a `table' element at the very beginning of a table (here
at the beginning of the first row), not a `table-row'.

Regards,
-- 
Nicolas Goaziou

Re: [PATCH] Fix test-org-element/cache

[Ihor Radchenko]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:
> I'm not sure to understand your report, but `org-element-at-point'
> should return a `table' element at the very beginning of a table (here
> at the beginning of the first row), not a `table-row'.

Sorry, I was probably not very clear. The test does the following:

1. Create initial file like:

#+begin_center
P0



  P1
  | a | b |
<point>| c | d |
#+end_center

Note that second line of the table is not indented.
Now, org-element-at-point returns table-row.

2. Modify the file to test org-element-cache

#+begin_center
P0

- item

  P1
  | a | b |
<point>| c | d |
#+end_center

Now, the first (indented) row of the table belongs to item.  The second
row becomes individual table and org-element-at-point returns table.

The patch asserted that org-element-at-point still returned a table row.

Best,
Ihor

Re: [PATCH] Fix test-org-element/cache

[Ihor Radchenko]

[-- Attachment #1: Type: text/plain, Size: 233 bytes --]

Bastien <bzg@gnu.org> writes:
> Applied in maint, thanks!

The patch also revealed an idea for another test.  The new test is
revealing existing bug on master.  Attached is the patch adding the test
with ":expected-result :failed".


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Add-a-test-for-known-bug-in-org-element-cache.patch --]
[-- Type: text/x-diff, Size: 1899 bytes --]

From d3a5a606c64ed1da0d1880f503c2461839dbfec9 Mon Sep 17 00:00:00 2001
Message-Id: <d3a5a606c64ed1da0d1880f503c2461839dbfec9.1621081760.git.yantar92@gmail.com>
From: Ihor Radchenko <yantar92@gmail.com>
Date: Sat, 15 May 2021 20:27:20 +0800
Subject: [PATCH] Add a test for known bug in org-element-cache

* testing/lisp/test-org-element.el (test-org-element/cache-bugs): A
new test documenting a known bug:

1. Create initial file like:

P0

  P1
  | a | b |
<point>| c | d |

Note that second line of the table is not indented.  Now,
org-element-at-point returns table-row.

2. Modify the file to test org-element-cache

P0

- item

  P1
  | a | b |
<point>| c | d |

Now, the first (indented) row of the table belongs to item.  The
second row should be an individual table and org-element-at-point
returns table.
---
 testing/lisp/test-org-element.el | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/testing/lisp/test-org-element.el b/testing/lisp/test-org-element.el
index 28e9ecd12..18991b5bc 100644
--- a/testing/lisp/test-org-element.el
+++ b/testing/lisp/test-org-element.el
@@ -3889,6 +3889,21 @@ (ert-deftest test-org-element/cache ()
 	    :end (org-element-property :parent (org-element-at-point)))
 	   (+ parent-end 3))))))
 
+(ert-deftest test-org-element/cache-bugs ()
+  "Test basic expectations and common pitfalls for cache."
+  :expected-result :failed
+  ;; Unindented second row of the table should not be re-parented by
+  ;; inserted item.
+  (should
+   (eq 'table
+       (let ((org-element-use-cache t))
+	 (org-test-with-temp-text
+	  "#+begin_center\nP0\n\n<point>\n\n  P1\n  | a | b |\n| c | d |\n#+end_center"
+	  (save-excursion (search-forward "| c |") (org-element-at-point))
+	  (insert "- item")
+	  (search-forward "| c |")
+	  (beginning-of-line)
+	  (org-element-type (org-element-at-point)))))))
 
 (provide 'test-org-element)
 
-- 
2.26.3

Re: [PATCH] Fix test-org-element/cache

[Ihor Radchenko]

> The patch asserted that org-element-at-point still returned a table row.
* The old test

Re: [PATCH] Fix test-org-element/cache

[Bastien]

Ihor Radchenko <yantar92@gmail.com> writes:

> Bastien <bzg@gnu.org> writes:
>> Applied in maint, thanks!
>
> The patch also revealed an idea for another test.  The new test is
> revealing existing bug on master.  Attached is the patch adding the test
> with ":expected-result :failed".

Applied in master, thanks.

-- 
 Bastien