Mercurial > hg > xemacs-beta
diff tests/automated/lisp-tests.el @ 5244:04811a268716
Be clearer in our error messages, #'canonicalize-inst-pair, #'canonicalize-spec
lisp/ChangeLog addition:
2010-08-15 Aidan Kehoe <kehoea@parhasard.net>
* specifier.el (canonicalize-inst-pair, canonicalize-spec):
If a specifier tag set is correct, but an instantiator is not in
an accepted format, don't error with the message "Invalid
specifier tag set".
Also, when we error, use error-symbols, for better structured
error handling and more ease when testing.
tests/ChangeLog addition:
2010-08-15 Aidan Kehoe <kehoea@parhasard.net>
* automated/lisp-tests.el:
(not, not, invalid-argument, invalid-argument):
Check that error messages from the image specifier instantiator
code are clearer than they used to be.
| author | Aidan Kehoe <kehoea@parhasard.net> |
|---|---|
| date | Sun, 15 Aug 2010 15:42:45 +0100 |
| parents | 808131ba4a57 |
| children | be436ac36ba4 308d34e9f07d |
line wrap: on
line diff
--- a/tests/automated/lisp-tests.el Sun Aug 15 13:29:10 2010 +0100 +++ b/tests/automated/lisp-tests.el Sun Aug 15 15:42:45 2010 +0100 @@ -2374,6 +2374,35 @@ (garbage-collect)))))) "checking we can amputate lists without crashing #'reduce") +(Assert (not (eq t (canonicalize-inst-list + `(((mswindows) . [string :data ,(make-string 20 0)]) + ((tty) . [string :data " "])) 'image t))) + "checking mswindows is always available as a specifier tag") + +(Assert (not (eq t (canonicalize-inst-list + `(((mswindows) . [nothing]) + ((tty) . [string :data " "])) + 'image t))) + "checking the correct syntax for a nothing image specifier works") + +(Check-Error-Message invalid-argument "^Invalid specifier tag set" + (canonicalize-inst-list + `(((,(gensym)) . [nothing]) + ((tty) . [string :data " "])) + 'image)) + +(Check-Error-Message invalid-argument "^Unrecognized keyword" + (canonicalize-inst-list + `(((mswindows) . [nothing :data "hi there"]) + ((tty) . [string :data " "])) 'image)) + +;; If we combine both the specifier inst list problems, we get the +;; unrecognized keyword error first, not the invalid specifier tag set +;; error. This is a little unintuitive; the specifier tag set thing is +;; processed first, and would seem to be more important. But anyone writing +;; code needs to solve both problems, it's reasonable to ask them to do it +;; in series rather than in parallel. + (when (featurep 'ratio) (Assert (not (eql '1/2 (read (prin1-to-string (intern "1/2"))))) "checking symbols with ratio-like names are printed distinctly")