Force empty lines non executable

[dvdoug]

To hopefully fix #799

This isn't quite the same concept as an "ignored line" so I've deliberately kept this implementation distinct from the static analysis code. It's done as a single pass at the end of execution for speed, although it also simplifies the code.

[codecov]

Codecov Report

Merging #805 into master will increase coverage by 0.04%.
The diff coverage is 100.00%.

@@             Coverage Diff              @@
##             master     #805      +/-   ##
============================================
+ Coverage     82.96%   83.01%   +0.04%     
- Complexity     1111     1118       +7     
============================================
  Files            60       60              
  Lines          3411     3421      +10     
============================================
+ Hits           2830     2840      +10     
  Misses          581      581              

Impacted Files	Coverage Δ	Complexity Δ
src/CodeCoverage.php	44.91% <100.00%> (+0.23%)	96.00 <0.00> (ø)
src/ProcessedCodeCoverageData.php	100.00% <100.00%> (ø)	58.00 <7.00> (+7.00)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 961fee7...1000da7. Read the comment docs.

[dvdoug]

The is_array isn't technically necessary, without it empty lines that are already non-executable are simply marked non-executable again. However I found it helpful for debugging to ensure that this only triggered on the cases I was expecting it to

[sebastianbergmann]

I've deliberately kept this implementation distinct from the static analysis code. It's done as a single pass at the end of execution for speed, although it also simplifies the code.

The downside of keeping this separate is that the result of this analysis does not get cached.

[lcobucci]

@dvdoug @sebastianbergmann it looks like this didn't solve the issue (or perhaps I'm doing something wrong?): lcobucci/jwt#380

[sebastianbergmann]

I am also seeing similar failures:

• https://github.com/sebastianbergmann/code-unit/runs/1061309910
• https://github.com/sebastianbergmann/comparator/runs/1061311241
• https://github.com/sebastianbergmann/global-state/runs/1061313373
• https://github.com/sebastianbergmann/object-enumerator/runs/1061314622
• https://github.com/sebastianbergmann/object-reflector/runs/1061315150
• https://github.com/sebastianbergmann/php-invoker/runs/1061316251
• https://github.com/sebastianbergmann/diff/runs/1061311774
• https://github.com/sebastianbergmann/cli-parser/runs/1061320703
• https://github.com/sebastianbergmann/lines-of-code/runs/1061319776
• https://github.com/sebastianbergmann/php-text-template/runs/1061318598
• https://github.com/sebastianbergmann/recursion-context/runs/1061317478
• https://github.com/sebastianbergmann/type/runs/1061317915
• https://github.com/sebastianbergmann/php-timer/runs/1061317036

[lcobucci]

@sebastianbergmann in v9.3.4 this component was relying on the driver to detect "dead code" (9.1.4...master#diff-b2fbc27b0e544bbffcdfba89b3fb1dd4L640), can the removal of that bit be cause of the problem?

[dvdoug]

The issue of the executable last line was originally raised in the context of reporting being incorrect, hence the implemention decision that:

It's done as a single pass at the end of execution for speed, although it also simplifies the code.

Those CI runs would appear to suggest that strict covers annotations means that it's not OK to do the analysis and adjustment as a one-off after the suite completes but requires it to be performed after each individual test invocation instead? That's annoying and will definitely want some caching 😢

I am happy to look into this more, but please note I am not likely to have time before next week

[sebastianbergmann]

I am happy to look into this more, but please note I am not likely to have time before next week

No worries, take your time. I will look into reverting the recent changes and roll a new release later today.