You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Typescript and Istanbul have issues (microsoft/TypeScript#24993) working together, but the nyc tool should support source maps if they have the correct paths.
Scope
make target
instrument with nyc
execute test bundle with instrumented main bundle
exclude node_modules and out, include src and test
gather test results
source maps
ext file for each bundle
correct path in bundle
istanbul/nyc load maps
include in npm bundle
upload results to code climate
token
tool image (if necessary)
pipeline job
set up codecov
repo setup
token
upload job
github required check
Use Case
Feature-driven tests are great, but coverage shows you dead code. To keep the bot lean as it grows, comprehensive tests and aggressive removal of unused code is vital.
Issues
Huge time sink, but very useful.
The text was updated successfully, but these errors were encountered:
Current test harness only includes files from src/ with some test coverage, leaving most of them out :( and skewing results too high :( . Either the --all argument to nyc or including src via webpack context should work, but node_modules and the out stubs still need to be excluded.
Until istanbuljs/nyc#953 is resolved, including sources leads to node_modules being instrumented, which runs indefinitely.
Working for now with the written tests and their included sources, but nyc itself is still including node_modules. Following up in their issues, using sed as a workaround to remove modules and webpack extras from lcov report before uploading. CI is set up with coverage, but numbers are high (not all src is included, only that imported by tests).
Summary
Typescript and Istanbul have issues (microsoft/TypeScript#24993) working together, but the
nyc
tool should support source maps if they have the correct paths.Scope
node_modules
andout
, includesrc
andtest
Use Case
Feature-driven tests are great, but coverage shows you dead code. To keep the bot lean as it grows, comprehensive tests and aggressive removal of unused code is vital.
Issues
Huge time sink, but very useful.
The text was updated successfully, but these errors were encountered: