[PATCH] D70105: [lit] Better/earlier errors for empty runs

[Julian Lettner via Phabricator via llvm-commits]

yln marked 6 inline comments as done.
yln added inline comments.


================
Comment at: llvm/utils/lit/lit/main.py:71
+            sys.stderr.write('Filter did not match any tests '
+                             '(of %d discovered).\n' % numTotalTests)
+            code = 0 if opts.allow_empty_runs else 2
----------------
jdenny wrote:
> When `--allow-empty-runs` is not already specified, could we have this diagnostic suggest it?  Otherwise, could we have it report that it was specified?
Good point. Will do.


================
Comment at: llvm/utils/lit/lit/run.py:48
         """
-        if not self.tests:
-            return 0.0
----------------
jdenny wrote:
> Would it be wortwhile to turn this guard into an assert rather than just removing it?
This function works for zero tests (although it then creates a process pool for no reason). It's just that you can't create a process pool with zero workers and we use `min(test_count, cpu_cores)` to determine the number of workers.  We have an assert for this above.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D70105/new/

https://reviews.llvm.org/D70105