feature(jest-runner): Pass the active mutant id through the env vars rather than config #2416
Conversation
|
|
||
| var __global_69fa48 = (function(g){ | ||
| if (g.__activeMutant__ === undefined && g.process && g.process.env && g.process.env.__STRYKER_ACTIVE_MUTANT__) { | ||
| g.__activeMutant__ = Number(g.process.env.__STRYKER_ACTIVE_MUTANT__); |
There was a problem hiding this comment.
This was the issue. As you can see, the process.env variable name was out-dated here (__ACTIVE_MUTANT__ -> __STRYKER_ACTIVE_MUTANT__).
The code here was instrumented with an older version of the instrumenter.
|
@nicojs Beta 4 does bring some really great performance improvements, thanks! The memory leak that I've noticed is still dragging down performance, but if I patch that out by restarting the process every 20 runs the performance improvement is easily 6x or more than compared to Beta 3. I'll look into bringing a project to you for use of real world testing |
|
Well, I spent all day, but I found out Jest just appears to be leaking for no reason at all. I've logged a bug report |
|
That's very interesting! I will do some checks as well and report my findings there. I've got a feeling that there should be a way to prevent memory leaks in Jest. Maybe some sort of |
This represents a rather massive difference in speed on our project. I can get propper numbers tomorrow, but from a quick look it's somewhere between 2x-4x at least.
It looks like the integration tests for jest-runner are failing. I'm not sure if it's because I missed something, or the integration tests missed something. It looks like the integration mutants aren't picking up the mutant id from the environment variables