While working on a dynamic context plugin I encountered an issue where contexts are not unique(I think it's related to subprocess coverage, but I haven't been able to create a small enough example to show here).

While we can argue that the plugin itself should be fixed, this looks like an innocuous change tocoveragepy which can prevent these misbehaving plugins from ruining most of the coverage data(not being covered).

Prevent misbehaving dynamic context plugins from ruining coverage

309f67e

s0undt3ch force-pushed thehotfix/context-on-conflict branch from69510e9 to309f67eCompare

January 4, 2021 07:22

Copy link

codecov-io commentedJan 4, 2021

Codecov Report

Merging#1088 (309f67e) intomaster (7ff93a9) willdecrease coverage by0.01%.
The diff coverage isn/a.

@@            Coverage Diff             @@##           master    #1088      +/-   ##==========================================- Coverage   91.28%   91.27%   -0.02%==========================================  Files          90       90                Lines       13358    13358                Branches     1493     1493              ==========================================- Hits        12194    12192       -2- Misses        953      954       +1- Partials      211      212       +1

Impacted Files	Coverage Δ
coverage/sqldata.py	`90.77% <ø> (ø)`
tests/test_oddball.py	`81.54% <0.00%> (-1.20%)`	⬇️

Copy link

Member

nedbat commentedJan 15, 2021

Do you have any more information about what caused the conflict? This is the type of change that can hide useful information about a problem to be fixed. I see that it would be safe to add, but I have qualms.

Copy link

Author

s0undt3ch commentedJan 15, 2021

Not really, no. The information I currently know is what's mentioned on the description, sub-processes coverage.
What the misbehaving plugin(mine) is doing, can be seenhere andhere.
Does it help understand the problem?