Tracy Walker wrote: > Last week bc and I discussed what needs to happen to get his set of > security regression tests fully automated. I'll attempt to automate any > test cases that require user interaction. I believe most everything else > will go into Mochikit. We need to start sorting the test cases into > those two buckets. We also need to determine where results should be > reported. > > bc, if you're able to get me a manual test case or two I can get started > with Eggplant. >
Setting followup-to: m.d.quality. Tracy, it might just be best to go directly to some of the bugs like <https://bugzilla.mozilla.org/show_bug.cgi?id=18653> and <https://bugzilla.mozilla.org/show_bug.cgi?id=37907> and working from the original testcases in the bugs rather than the "mfTest" based versions I had scrapped from the bugs in my previous life. All of the bugs I had previously created "mfTest" tests for, have now had their in-testsuite flags set to ? and should be easily queryable. The real question I wanted to ask folks is how do we handle these tests. For tests which are passing in released products, I don't see any problem in hosting the tests and executing them in public. For new tests which do not pass in released products, what do we do? Host them separately behind the firewall with access restricted to the security group? Do we need to have two testing infrastructures? one public and one private? Then we have the issue between the mochikit-style automated tests and the user-interactive tests which are to be automated using eggplant. Do people have any opinions they would like to share? /bc _______________________________________________ dev-security mailing list [email protected] https://lists.mozilla.org/listinfo/dev-security
