summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorGravatar Joey Hess <joey@kitenet.net>2014-07-04 18:07:41 -0400
committerGravatar Joey Hess <joey@kitenet.net>2014-07-04 18:07:41 -0400
commit98986ee70bf44fa221120c4f31a8eaf11a35d8e6 (patch)
treee7c7cadf3ee75464d029bc1e3445c6a9f1725bde
parent0bcf061abf5f2a63dbcd71290863615f991b5158 (diff)
devblog
-rw-r--r--doc/devblog/day_190__fun_fixes.mdwn13
1 files changed, 13 insertions, 0 deletions
diff --git a/doc/devblog/day_190__fun_fixes.mdwn b/doc/devblog/day_190__fun_fixes.mdwn
new file mode 100644
index 000000000..d3895df1f
--- /dev/null
+++ b/doc/devblog/day_190__fun_fixes.mdwn
@@ -0,0 +1,13 @@
+Spent the morning improving behavior when `commit.gpgsign` is set.
+Now git-annex will let gpg sign commits that are made when eg, manually
+running `git annex sync`, but not commits implicitly made to the git-annex
+branch. And any commits made by the assistant are not gpg signed. This was
+slightly tricky, since lots of different places in git-annex ran `git
+commit`, `git merge` and similar.
+
+Then got back to a test I left running over vacation, that added millions
+of files to a git annex repo. This was able to reproduce a problem where
+`git annex add` blew the stack and crashed at the end. There turned out to
+be two different memory issues, one was in git-annex and the other is in
+Haskell's core `getDirectoryContents`. Was able to entirely fix it,
+eventually.