Some brittleness in tests

[phst]

For example on Ubuntu 12.04:

$ LC_ALL=C make test SELECTOR='(rx bos "test-markdown-font-lock/hidden-urls-")'
emacs -Q --batch  \
	    --eval '(setq byte-compile-warnings (quote (not cl-functions)))' \
	    --eval '(setq byte-compile-error-on-warn t)' \
	    -l ../markdown-mode.el \
	    -f batch-byte-compile ../markdown-mode.el markdown-test.el
emacs -Q --batch  -l checkdoc-batch.el 2>&1 | \
	    grep -E "markdown-mode.el:[1-9]+" && exit 1 || exit 0
emacs -Q --batch  \
	    --eval '(setq byte-compile-warnings (quote (not cl-functions)))' \
	    -l ert -l ../markdown-mode.elc -l markdown-test.elc \
	    --eval '(ert-run-tests-batch-and-exit (rx bos "test-markdown-font-lock/hidden-urls-"))'
Running 2 tests (2017-08-31 21:21:06+0200)
Test test-markdown-font-lock/hidden-urls-inline backtrace:
  #[nil "\306�	\307P�\310 311\216\312\313!�r\fq\210\314\216\315\n!\2
  ert--run-test-internal([cl-struct-ert--test-execution-info [cl-struc
  ert-run-test([cl-struct-ert-test test-markdown-font-lock/hidden-urls
  ert-run-or-rerun-test([cl-struct-ert--stats "\\`test-markdown-font-l
  ert-run-tests("\\`test-markdown-font-lock/hidden-urls-" #[385 "\306�
  ert-run-tests-batch("\\`test-markdown-font-lock/hidden-urls-")
  ert-run-tests-batch-and-exit("\\`test-markdown-font-lock/hidden-urls
  eval((ert-run-tests-batch-and-exit (rx bos "test-markdown-font-lock/
  command-line-1(("--eval" "(setq byte-compile-warnings (quote (not cl
  command-line()
  normal-top-level()
Test test-markdown-font-lock/hidden-urls-inline condition:
    (ert-test-failed
     ((should
       (equal '...
	(get-text-property 932 ...)))
      :form
      (equal
       ((26 . 8734))
       ((26 . 35)))
      :value nil :explanation
      (list-elt 0
		(cdr
		 (different-atoms ... ...)))))
   FAILED  1/2  test-markdown-font-lock/hidden-urls-inline
Test test-markdown-font-lock/hidden-urls-reference backtrace:
  #[nil "\306�\307 �\310\216\311\312!�r\nq\210\313\216\314	p\306#\210\
  ert--run-test-internal([cl-struct-ert--test-execution-info [cl-struc
  ert-run-test([cl-struct-ert-test test-markdown-font-lock/hidden-urls
  ert-run-or-rerun-test([cl-struct-ert--stats "\\`test-markdown-font-l
  ert-run-tests("\\`test-markdown-font-lock/hidden-urls-" #[385 "\306�
  ert-run-tests-batch("\\`test-markdown-font-lock/hidden-urls-")
  ert-run-tests-batch-and-exit("\\`test-markdown-font-lock/hidden-urls
  eval((ert-run-tests-batch-and-exit (rx bos "test-markdown-font-lock/
  command-line-1(("--eval" "(setq byte-compile-warnings (quote (not cl
  command-line()
  normal-top-level()
Test test-markdown-font-lock/hidden-urls-reference condition:
    (ert-test-failed
     ((should
       (equal '...
        (get-text-property 8 ...)))
      :form
      (equal
       ((20 . 8734))
       ((20 . 35)))
      :value nil :explanation
      (list-elt 0
                (cdr
                 (different-atoms ... ...)))))
   FAILED  2/2  test-markdown-font-lock/hidden-urls-reference

Ran 2 tests, 0 results as expected, 2 unexpected (2017-08-31 21:21:06+0200)

2 unexpected results:
   FAILED  test-markdown-font-lock/hidden-urls-inline
   FAILED  test-markdown-font-lock/hidden-urls-reference

make: *** [test] Error 1

This is because with LC_ALL=C Emacs falls back to # for the URL compose character. Given

But the value of the property itself is completely internal to Emacs and should not be manipulated directly by, for instance, put-text-property.

(https://www.gnu.org/software/emacs/manual/html_node/elisp/Special-Properties.html for the composition property), maybe the tests should only check whether the property is nil.

[jrblevin]

I had not seen this; thanks for letting me know.