On 10/27/19 7:44 PM, brian m. carlson wrote:

On 2019-10-22 at 11:45:17, Vegard Nossum wrote:

quoted

diff --git a/t/t5100/meta-meta0001 b/t/t5100/meta-meta0001
new file mode 100644
index 0000000000..1db7e9f715
--- /dev/null
+++ b/t/t5100/meta-meta0001

@@ -0,0 +1,23 @@
+commit 763f9b1cfd69ade5e22dcdcdc35d144697675a93
+tree 43a6b213a2891b5a863775771cab0c0dba3730dc
+parent 108b97dc372828f0e72e56bbb40cae8e1e83ece6
+author Vegard Nossum <vegard.nossum@oracle.com> 1570284959 +0200
+committer Vegard Nossum <vegard.nossum@oracle.com> 1571599909 +0200
+gpgsig -----BEGIN PGP SIGNATURE-----
+ Version: GnuPG v1
+
+ iQIcBAABAgAGBQJdrLYmAAoJEAvO9Nj+mLpY5zIQAJkdnnZMrCVme64r43M/KMm0
+ W1fmdeXiIMrI7i0McBdAsQ/KQ5yD4HBvaJWCyI0/g6IeLgVBf9//9xq4Y32iqsKn
+ XRut2Pk3H3Az0WfUlpLpDJUgzz7er8t/glaKnESb94XR/ac59tEELbePh+bXsFLH
+ 3+v8Y78zeHJd6ZLKrKmLyq/9ZaJQR+9xmGdKzZdnwM+8seE4aOhM1VtA8ik68Tn6
+ Tbaofp1jbsXcyY4nBG9GxK14wnb/8OZmQlH4J40LsZT4KDWQNWighGig8ude7SJa
+ 6FJXWJPLfOB6r2ThiJUnrf/UXeHbvYUWITiYoWOxEVb6c7RfNLqDbbGF9VQfTD0n
+ SEFO5NqOs6KofaYzALprtgpMrqksRHeLc7Ouh9xgLyLZLx/669I9lo8M1aQ7RJMy
+ V1KDG6sYbFrgy4gQ/4xqXj3NpBmMb/VcjOnCj3j040wo8q7hlpzb6ev5lcqAFEuP
+ y1owwhljMjqAdGIBw6sLVn2on+6gEQuIjbkoapBktPDw7xEpOKe9rzTGcWRRyANs
+ Z+pMWbn8c6TKonokNjERy0iPnu2t2j8x1YpqKdjY+oq8ApNZFMlU1U+UtXFfuLw5
+ ZKR5DtmXxWzvd+nKBenjzXyOt33v5eq4I/WMfATauXBgFu75mbfiKIFVD5VeMfxq
+ DoiIqvLtW+DDUkH99zXm
+ =2APb
+ -----END PGP SIGNATURE-----

First, let me say that I'm pleased to see this series.

It would be nice if we could use the test user and test keys, since this
test data isn't going to work for SHA-256 and I'll need to generate
suitable test data for this test as part of porting it.  I won't be able
to forge a signature using your personal key.

If you wanted to generate that data yourself instead, you're welcome to
rebase your changes onto the transition-stage-4 branch from
https://github.com/bk2204/git.git and run the testsuite with
GIT_TEST_DEFAULT_HASH=sha256 to see where it fails.

Hi!

Thanks for the feedback.

I rebased on your branch and ran the tests without any problems.

t5100 is fine for me -- the file above is only used for testing the mail
splitting algorithm, so the choice of hashing algorithm should be
irrelevant for the test. That's also why I chose to use a personal key
for the signature, the other tests also use pregenerated emails to avoid
testing the email generation in those tests. (It could be useful to
include those "historical" test cases, which should still work in case
the format ever changes again in the future.)

The other tests generate the commits/emails within the test. Are you
sure you resolved the conflict in t4150 correctly (i.e. leaving out the
gitattributes test at the end)?


Vegard