coverage combine when not in parallel mode wipes .coverage information (regression from 4.1)

[nedbat]

Originally reported by Anthony Sottile (Bitbucket: asottile, GitHub: asottile)

---

Not sure if this intentionally or accidentally regressed. Prior to 4.2, running coverage combine was a noop if the coverage was already combined. Now it wipes the data.

Simple reproduction

echo 'x = 5' > test.py
coverage run test.py
coverage report --include test.py
coverage combine
coverage report --include test.py

4.1

$ echo 'x = 5' > test.py
$ coverage run test.py
$ coverage report --include test.py
Name      Stmts   Miss  Cover
-----------------------------
test.py       1      0   100%
$ coverage combine
$ coverage report --include test.py
Name      Stmts   Miss  Cover
-----------------------------
test.py       1      0   100%

4.2

$ echo 'x = 5' > test.py
$ coverage run test.py
$ coverage report --include test.py
Name      Stmts   Miss  Cover
-----------------------------
test.py       1      0   100%
$ coverage combine
$ coverage report --include test.py
Name    Stmts   Miss  Cover
---------------------------
No data to report.

---

• Bitbucket: https://bitbucket.org/ned/coveragepy/issue/525

[nedbat]

Yes, this was a regression. It's already fixed on the tip.

[nedbat]

Fixed in 34ae57c51cfb3b2cd76725acc92aaed57bad1e79 (bb) (two months ago, yikes, I should release it!)

[nedbat]

Original comment by Anthony Sottile (Bitbucket: asottile, GitHub: asottile)

---

Neat, thanks again @nedbat!

[nedbat]

This fix was released as part of Coverage.py 4.3.