Sorry, no issue, no test script to reproduce. Still think this is correct fix.

1 patch for repository http://darcs.net/screened:

patch 87d44f25f7c4c78c36832cdab55ecc2bd47c1bbf
Author: Ben Franksen <benjamin.franksen@helmholtz-berlin.de>
Date:   Wed Nov 21 18:45:41 CET 2018
  * fix in darcs check: withTempDir -> withDelayedDir
  
  Every now and again darcs check fails for me because it tries to read a
  patch from the temporary directory which doesn't exist any more.

Attachments

• fix-in-darcs-check_-withtempdir-__-withdelayeddir.dpatch
• patch-preview.txt
• unnamed

Is this working for you in practice? It seems a bit strange because 
from the code all the work should be done inside the "withTempDir"
call. But happy to apply if it does make a difference (perhaps
with a comment in the code explaining why it's that way).

> Is this working for you in practice? It seems a bit strange because 
> from the code all the work should be done inside the "withTempDir"
> call. But happy to apply if it does make a difference (perhaps
> with a comment in the code explaining why it's that way).

It does and I never again had the problem I mentioned. I guess the
reason this failed occasionally may be because of laziness somewhere
inside the action?

So replayRepository' may return a new pristine that is read from the 
temporary location and reading a Tree is done using lazy ByteStrings 
(for file contents). Then we check if there is a difference to our 
stored pristine, but when there are differences the check may 
terminate early and not all of the new pristine was read/evaluated.

I see, thanks. Could you add a comment to that effect?

Following up on review.

1 patch for repository http://darcs.net/screened:

patch 3efbb66b2a35f29cc384613017f625ee9eab0a54
Author: Ben Franksen <ben.franksen@online.de>
Date:   Fri Jun 14 13:58:01 CEST 2019
  * add comment to explain use of withDelayedDir

Attachments

• add-comment-to-explain-use-of-withdelayeddir.dpatch
• patch-preview.txt
• unnamed