Code coverage check incorrectly compares to previous build

[tomasMizera]

It seems that code coverage check currently compares coverage to previous build on the same branch. It is fine for master branch, but in feature branches this is wrong due to scenarios like this:

Let's say initial code coverage is 60%,

• create commit A (without autotests), coverage drops to 59.5% - check fails (✔️ )
• create commit B (without autotests), coverage drops to 59% - check fails (✔️ )
• create commit C (without autotests, but only fix layout), coverage is still 59% - check succeeds because it compares to previous build (❌ ) .. However, overall coverage still dropped by 1%.

I guess we should always compare to last master build?

[PeterPetrik]

yep, should compare to master....

[tomasMizera]

See https://github.com/lemurheavy/coveralls-public/issues/1167