[Fix #8629] Calculate rubocop_checksum by calculating a crc32 for eac…

…h file rather than using mtime.

File.mtime is faster, but inconsistent in CI, because in each CI build the mtime changes, which means that the rubocop cache implicitly cannot be used as-is. It's obviously slower to calculate a hash for each file, but this is still more performant (both in memory consumption and iterations per second) than the original.

CHANGELOG.md

@@ -56,6 +56,7 @@
 * [#8517](https://github.com/rubocop-hq/rubocop/pull/8517): Make `Style/HashTransformKeys` and `Style/HashTransformValues` aware of `to_h` with block. ([@eugeneius][])
 * [#8529](https://github.com/rubocop-hq/rubocop/pull/8529): Mark `Lint/FrozenStringLiteralComment` as `Safe`, but with unsafe auto-correction. ([@marcandre][])
 * [#8602](https://github.com/rubocop-hq/rubocop/pull/8602): Fix usage of `to_enum(:scan, regexp)` to work on TruffleRuby. ([@jaimerave][])
+* [#8629](https://github.com/rubocop-hq/rubocop/pull/8629): Fix the cache being reusable in CI by using crc32 to calculate file hashes rather than `mtime`, which changes each CI build. ([@dvandersluis][])
 
 ## 0.89.1 (2020-08-10)
 

lib/rubocop/result_cache.rb

@@ -3,6 +3,7 @@
 require 'digest/sha1'
 require 'find'
 require 'etc'
+require 'zlib'
 
 module RuboCop
   # Provides functionality for caching rubocop runs.
@@ -171,7 +172,7 @@ def rubocop_checksum
           rubocop_extra_features
             .select { |path| File.file?(path) }
             .sort!
-            .each { |path| digest << File.mtime(path).to_s }
+            .each { |path| digest << Zlib.crc32(IO.read(path)).to_s }
           digest << RuboCop::Version::STRING << RuboCop::AST::Version::STRING
           digest.hexdigest
         end