Re: RFC: feature/auth-source-pass branch integrating auth-source with the pass secret manager

[Noam Postavsky]

On Mon, Mar 27, 2017 at 1:45 PM, Ted Zlatanov <address@hidden> wrote:
> Please comment on the code, if you have any questions or concerns.

+(defun auth-source-pass--remove-directory-name (name)
+  "Remove directories from NAME.
+E.g., if NAME is \"foo/bar\", return \"bar\"."
+  (replace-regexp-in-string ".*/" "" name))

This could be replaced with `file-name-nondirectory', I think.

+(defun auth-source-pass--should-have-message-containing (regexp)
+  "Assert that at least one `auth-source-do-debug` matched REGEXP."
+  (should (seq-find (lambda (message)
+                      (string-match regexp message))
+                    auth-source-pass--debug-log)))

The reliance on debug prints for the tests seems a bit strange to me
(I don't really have a better suggestion though).

+(defmacro auth-source-pass--deftest (name arglist store &rest body)
+  "Define a new ert-test NAME with ARGLIST using STORE as password-store.
+BODY is a sequence of instructions that will be evaluated.
+
+This macro overrides `auth-source-pass-parse-entry' and
`auth-source-pass-entries' to
+test code without touching the file system."

I would suggest rather defining auth-source-pass--with-store, and
leaving the test defining part to ert-deftest.

So that, e.g.,

(auth-source-pass--deftest auth-source-pass-find-match-matching-at-entry-name ()
           '(("foo"))
  (should (equal (auth-source-pass--find-match "foo" nil)
                 "foo")))

becomes

(ert-deftest auth-source-pass-find-match-matching-at-entry-name ()
  (auth-source-pass--with-store '(("foo"))
    (should (equal (auth-source-pass--find-match "foo" nil)
                   "foo"))))