Closed Bug 846113 Opened 7 years ago Closed 3 years ago

Tracking: "API complete" testing of newly added APIs

Categories

(Core :: General, defect)

x86_64
Linux
defect
Not set

Tracking

()

RESOLVED INCOMPLETE

People

(Reporter: cjones, Unassigned)

References

Details

We have new APIs with very good test coverage, but on the whole we haven't done a great job with this.  Part of the problem is economics: with hundreds/thousands of blocking bugs, it's hard to justify the cost of spending a lot of time spinning up a test harness for an API with a small state space, that's being tested and hacked on frequently by developers.

But as we move forwards and developers start to work on new features, regressions become a constant, increasing cost.  And in particular, for big new projects like bug 761935, we have no hope of not Breaking The World without automated testing.

A problem with testing some of the new APIs we've created is that writing the /first/ test is often very difficult.  Test harnesses rarely fit new uses 100%, so some hard problems have to be solved to clear a path for the first test.  After that, additional tests are much easier, since they can usually follow the example of the first.

The project of this meta-bug is to "clear the paths" for all the new APIs; solve the problems that block having a first test written for each.  After that, we can fill state-space coverage using normal processes, as we go along.  For example, using branch coverage metrics.

To do this we'll need to audit the new APIs we've added.
Status: NEW → RESOLVED
Closed: 3 years ago
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.