3 patches for repository http://darcs.net/screened:

patch ff1fb3d035d1af0320e1d06c3c20c93f18493195
Author: Ben Franksen <ben.franksen@online.de>
Date:   Fri Dec 14 16:20:04 CET 2018
  * several fixes in tests/failed-amend-should-not-break-repo.sh
  
  The darcs invocation here failed for the wrong reason (it expected more
  interactive input). Furthermore, issue1406 is no longer an issue as it
  concerns only the old-fashioned un-hashed repo format; the test now checks
  that if amend --test fails, then output of darcs log before and after the
  amend is equal.

patch 38fd0423cf79cb843d309d114f295c2f697fae61
Author: Ben Franksen <ben.franksen@online.de>
Date:   Fri Dec 14 16:54:05 CET 2018
  * in some test scripts, remove test repo before creating it (and not at the end)

patch 53e837de2ef5deb5e19147012e7d7ebf68c97c83
Author: Ben Franksen <ben.franksen@online.de>
Date:   Fri Dec 14 16:55:54 CET 2018
  * in test for issue2271, reset permissions more reliably when done

Attachments

• patch-preview.txt
• several-fixes-in-tests_failed_amend_should_not_break_repo_sh.dpatch
• unnamed

>  * several fixes in tests/failed-amend-should-not-break-repo.sh
>  * in some test scripts, remove test repo before creating it (and not at the end)
>  * in test for issue2271, reset permissions more reliably when done

All OK

For the last one, I spent a little while trying to make sure it 
works on Windows. The test still passes fine, but I'm really
confused about how our tests and darcs interact with Windows
permissions: I couldn't see any difference in the permissions
for the patch_index folder before and after the chmod -w, and
could remove the folder myself from the command-line fine.
Yet darcs still correctly fails/doesn't fail depending on whether
it is run.