Test failure for julia--test-end-of-defun-nested-2

[simonlegrand]

I just tried to run the tests for 7a8c868, with Emacs-29.1, and I got the following error:

...
Test julia--test-end-of-defun-nested-2 backtrace:
  ert-fail(((should (eq (point) (if (stringp "return fact(x)[ \n]+end"
  (if (unwind-protect (setq value-657 (apply fn-655 args-656)) (setq f
  (let (form-description-659) (if (unwind-protect (setq value-657 (app
  (let ((value-657 'ert-form-evaluation-aborted-658)) (let (form-descr
  (let* ((fn-655 #'eq) (args-656 (condition-case err (let ((signal-hoo
  (progn (julia-mode) (insert "function f(x)\nfunction fact(n)\nif n =
  (unwind-protect (progn (julia-mode) (insert "function f(x)\nfunction
  (save-current-buffer (set-buffer temp-buffer) (unwind-protect (progn
  (let ((temp-buffer (generate-new-buffer " *temp*" t))) (save-current
  (lambda nil (let ((temp-buffer (generate-new-buffer " *temp*" t))) (
  ert--run-test-internal(#s(ert--test-execution-info :test #s(ert-test
  ert-run-test(#s(ert-test :name julia--test-end-of-defun-nested-2 :do
  ert-run-or-rerun-test(#s(ert--stats :selector t :tests ... :test-map
  ert-run-tests(t #f(compiled-function (event-type &rest event-args) #
  ert-run-tests-batch(nil)
  ert-run-tests-batch-and-exit()
  command-line-1(("-L" "." "-l" "ert" "-l" "julia-mode-tests.el" "-f" 
  command-line()
  normal-top-level()
Test julia--test-end-of-defun-nested-2 condition:
    (ert-test-failed
     ((should
       (eq
	(point)
	(if ... ... "return fact(x)[ \n]+end")))
      :form
      (eq 141 163)
      :value nil))
   FAILED  19/65  julia--test-end-of-defun-nested-2 (0.001256 sec) at julia-mode-tests.el:887
...

I tried with Emacs-28.2 and it works fine.

Cheers,

Simon

[rodrigomorales1]

This problem seems to be the cause of the problem for this behavior in Guix

$ guix package -i emacs-ess

The following package will be installed:
   emacs-ess 18.10.2-0.24da603

The following derivations will be built:
  /gnu/store/fr1fp22rpdjc7k4my8b5bdvqpvf14s8y-emacs-julia-mode-0.4.drv
  /gnu/store/y6gn2gkvnawzkf5n79ryswqcx6r809rn-emacs-ess-18.10.2-0.24da603.drv

building /gnu/store/fr1fp22rpdjc7k4my8b5bdvqpvf14s8y-emacs-julia-mode-0.4.drv...
| 'check' phasebuilder for `/gnu/store/fr1fp22rpdjc7k4my8b5bdvqpvf14s8y-emacs-julia-mode-0.4.drv' failed with exit code 1
build of /gnu/store/fr1fp22rpdjc7k4my8b5bdvqpvf14s8y-emacs-julia-mode-0.4.drv failed
View build log at '/var/log/guix/drvs/fr/1fp22rpdjc7k4my8b5bdvqpvf14s8y-emacs-julia-mode-0.4.drv.gz'.
guix package: error: build of `/gnu/store/fr1fp22rpdjc7k4my8b5bdvqpvf14s8y-emacs-julia-mode-0.4.drv' failed

$ echo $?

1

Here's the relevant part of the file /var/log/guix/drvs/fr/1fp22rpdjc7k4my8b5bdvqpvf14s8y-emacs-julia-mode-0.4.drv.gz' (complete file here)

Indenting region... 
Indenting region...done
Test julia--test-end-of-defun-nested-2 backtrace:
  signal(ert-test-failed (((should (eq (point) (if (stringp "return fa
  ert-fail(((should (eq (point) (if (stringp "return fact(x)[ \n]+end"
  (if (unwind-protect (setq value-247 (apply fn-245 args-246)) (setq f
  (let (form-description-249) (if (unwind-protect (setq value-247 (app
  (let ((value-247 'ert-form-evaluation-aborted-248)) (let (form-descr
  (let* ((fn-245 #'eq) (args-246 (condition-case err (let ((signal-hoo
  (progn (julia-mode) (insert "function f(x)\nfunction fact(n)\nif n =
  (unwind-protect (progn (julia-mode) (insert "function f(x)\nfunction
  (save-current-buffer (set-buffer temp-buffer) (unwind-protect (progn
  (let ((temp-buffer (generate-new-buffer " *temp*" t))) (save-current
  (lambda nil (let ((temp-buffer (generate-new-buffer " *temp*" t))) (
  ert--run-test-internal(#s(ert--test-execution-info :test #s(ert-test
  ert-run-test(#s(ert-test :name julia--test-end-of-defun-nested-2 :do
  ert-run-or-rerun-test(#s(ert--stats :selector t :tests ... :test-map
  ert-run-tests(t #f(compiled-function (event-type &rest event-args) #
  ert-run-tests-batch(nil)
  ert-run-tests-batch-and-exit()
  command-line-1(("-l" "julia-mode-tests.el" "-f" "ert-run-tests-batch
  command-line()
  normal-top-level()
Test julia--test-end-of-defun-nested-2 condition:
    (ert-test-failed
     ((should
       (eq
	(point)
	(if ... ... "return fact(x)[ \n]+end")))
      :form
      (eq 141 163)
      :value nil))
   FAILED  14/43  julia--test-end-of-defun-nested-2 (0.000723 sec) at julia-mode-tests.el:584
   passed  15/43  julia--test-escaped-strings-dont-terminate-string (0.000403 sec)
   passed  16/43  julia--test-function-assignment-font-locking (0.000419 sec)
Indenting region... 
Indenting region...done

[zimoun]

@simonlegrand I confim the bug appears with Emacs 29.1 and not with 28.2. What happens? Consider the Julia code used by the test:

function f(x)
    function fact(n)
        if n == 0
            return 1
        else
            return n * fact(n-1)
        end
    end
    return fact(x)
end

Now, I use the symbol [X] to mark where the current point is.

function f(x)
    function fact(n)
        if n == 0[X]
            return 1
        else
            return n * fact(n-1)
        end
    end
    return fact(x)
end

The test julia--test-end-of-defun-nested-2 applies C-u 2 C-M-e, it results:

function f(x)
    function fact(n)
        if n == 0
            return 1
        else
            return n * fact(n-1)
        end
    end
[x]    return fact(x)
end

which is incorrect. It should be at the end. Therefore, julia-end-of-defun is buggy.

I guess the bug comes from this new snippet in 29.1, lisp/emacs-lisp/lisp.el from core:

        ;; At this point, point either didn't move (because we started
        ;; in between two defun's), or is at the end of a defun
        ;; (because we started in the middle of a defun).
        (unless (zerop arg)
          (when (setq success (beginning-of-defun-raw (- arg)))
            (funcall end-of-defun-function))))

that replaced this old snippet in 28.x:

        (unless (zerop arg)
          (beginning-of-defun-raw (- arg))
	  (setq beg (point))
          (funcall end-of-defun-function))))

[zimoun]

The test julia--test-end-of-defun-nested-2 applies C-u 2 C-M-e, it results:
[...]
which is incorrect. It should be at the end. Therefore, julia-end-of-defun is buggy.

Hum, I do not know... I observe the same behavior with an equivalent Python code using python-mode.

[simonlegrand]

Thanks for the explanations!

Hum, I do not know... I observe the same behavior with an equivalent Python code using python-mode.

Doesn't it make sens since python-mode is also using end-of-defun-function?

[zimoun]

Reported to upstream Emacs: #66763.

[tpapp]

While waiting for Emacs to address this, I refactored our tests in #209 so they no longer error. Closing this, but if anyone wants it reopened please comment.