Tracking Issue: Split up our UI-tests into smaller parts

[oli-obk]

Changing files like tests/ui/methods.rs causes a lot of diff noise. I think the ui output of every file should be kept below 100 lines preferrably and below 200 as a hard limit.

The files would best be split up to only check one or two lints

Current status:

$ wc -l tests/ui/*.stderr | sort -nr | head -n 15
   272 tests/ui/cognitive_complexity.stderr
   260 tests/ui/collapsible_if.stderr
   259 tests/ui/non_copy_const.stderr
   258 tests/ui/missing-doc.stderr
   247 tests/ui/if_same_then_else.stderr
   246 tests/ui/transmute.stderr
   245 tests/ui/match_same_arms.stderr
   224 tests/ui/needless_range_loop.stderr
   220 tests/ui/drop_forget_ref.stderr
   214 tests/ui/use_self.stderr
   214 tests/ui/eq_op.stderr
   213 tests/ui/indexing_slicing.stderr
   210 tests/ui/option_map_unit_fn.stderr
   203 tests/ui/booleans.stderr

[llogiq]

Perhaps we should also add a check to our test suite so we get a warning if a test gets too big?

[oli-obk]

We'd need to patch compiletest to do that nicely. Which reminds me... I wanted to make compiletest work on stable ^^

[mcarton]

👍 UI tests can be really hard to check for updates when they are too long

[llogiq]

Looking at wc output, I think we should set the limit at 300 lines of stderr output:

   464 tests/ui/cast.stderr
   582 tests/ui/for_loop.stderr
   742 tests/ui/methods.stderr

[phansch]

Currently looks like this:

$ wc -l tests/ui/*.stderr | sort -nr | head -n 15
   421 tests/ui/for_loop.stderr
   405 tests/ui/methods.stderr
   392 tests/ui/copies.stderr
   313 tests/ui/checked_unwrap.stderr
   309 tests/ui/indexing_slicing.stderr
   282 tests/ui/matches.stderr
   275 tests/ui/non_copy_const.stderr
   272 tests/ui/cyclomatic_complexity.stderr
   260 tests/ui/collapsible_if.stderr
   258 tests/ui/missing-doc.stderr
   246 tests/ui/transmute.stderr
   220 tests/ui/drop_forget_ref.stderr
   218 tests/ui/replace_consts.stderr
   218 tests/ui/literals.stderr