patch, 2 partsSubject: [PATCH 0/2] v2.53.0-rc0 test failures on cygwin

This cycle brings a t9700 test failure on cygwin, see patch #1. The second patch was owed from the last cycle, sorry about not getting to it sooner, but it is _still_ a puzzle. ;)

I have tested these patches on Linux without issue. On cygwin the t0610 test passes with '--stress' which it failed earlier and the t9700 test has only been tested by hand.

Directly after sending these patches, I will start a '-j8' run of the full testsuite on cygwin and let you know if it fails (it should not!). ;)

[When I run the testsuite like this on cygwin I can't use the laptop for anything else and the keyboard gets _very_ toasty!]

Thanks

Ramsay Jones

Ramsay Jones (2):
  t9700/test.pl: fix path type expectation on cygwin
  t0610-reftable-basics: mitigate a flaky test on cygwin

 t/t0610-reftable-basics.sh | 2 +-
 t/t9700/test.pl            | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

On 16/01/2026 8:39 pm, Ramsay Jones wrote:

Just for avoidance of doubt, this test passed without issue:

  $ ./git version
  git version 2.53.0.rc0.2.gc56c9e3a41
  $ 

[ie. v2.53.0-rc0 + these two patches]

  $ tail test-out1
  All tests successful.
  Files=1028, Tests=32260, 2945 wallclock secs (22.75 usr 64.42 sys + 7655.90 cusr 10222.07 csys = 17965.14 CPU)
  Result: PASS
  make clean-except-prove-cache
  make[2]: Entering directory '/home/ramsay/git/t'
  rm -f -r 'chainlinttmp'
  rm -f -r 'trash directory'.*
  rm -f -r valgrind/bin
  make[2]: Leaving directory '/home/ramsay/git/t'
  make[1]: Leaving directory '/home/ramsay/git/t'
  $ 

Now I can try a lower '-j' value to find a happy medium (although no '-j' argument leads to approx. 4 hours for a test run, which is already better than 6+ hours for my old laptop).

Thanks.

ATB, Ramsay Jones