327163 - Various tests fail in TestCookie

Status: RESOLVED DUPLICATE of bug 338332

(Core :: Networking: Cookies, defect)

Description

[Christian :Biesinger (don't email me, ping me on IRC)]

$ ./run-mozilla.sh ./TestCookie 2>&1 |grep FAIL
*** tests 5 6 FAILED!
*** tests 4 8 9 10 FAILED!
*** tests 4 5 6 9 10 12 FAILED!
*** tests 0 1 2 4 6 10 12 FAILED!
*** tests 8 FAILED!
*** tests 2 4 5 7 11 12 13 FAILED!
*** tests 1 4 FAILED!
*** tests 0 1 2 4 5 6 FAILED!
*** Result: TEST(S) FAILED!

Either the tests show a bug, or the test is wrong...

Comment 1

[[e-mail]]

i haven't looked at the tests in a while but when i wrote them, they used to all pass.

Comment 2

[Boris Zbarsky [:bzbarsky]]

We're failing because cookie.get() is "" instead of NULL.  And that's not that surprising -- the glue nsCString version is not at all the same things as nsXPIDLCString.  That's fallout from bug 315087, which changed the string type.

Comment 3

[[e-mail]]

i wonder if javascript requires this NULL behavior, or if we can just rework the consumers to deal with the empty string...

Comment 4

[Boris Zbarsky [:bzbarsky]]

What does JS have to do with this?  TestCookie is a C++ file....

Comment 5

[[e-mail]]

er, sorry. i thought you meant nsCookieService had been changed to use CStrings; nevermind.

Comment 6

[Boris Zbarsky [:bzbarsky]]

OK, so we either need to change the test to expect something different, or we need to change the test to use a different string type, or we need to change the behavior of nsCString here.

I'm going to assume the first of these is what we want and will attach a patch.

Comment 7

[Boris Zbarsky [:bzbarsky]]

Attachment: Simplest possible patch

With this, I pass the tests.

Comment 8

[Darin Fisher]

this is a duplicate of bug 338332, which davel already patched weeks ago, but his patch just never made it into the tree :-(

*** This bug has been marked as a duplicate of 338332 ***