update per other comment
authorJoey Hess <joeyh@joeyh.name>
Wed, 20 Oct 2021 17:12:16 +0000 (13:12 -0400)
committerJoey Hess <joeyh@joeyh.name>
Wed, 20 Oct 2021 17:12:16 +0000 (13:12 -0400)
doc/bugs/WSL_adjusted_braches__58___smudge_fails_with_sqlite_thread_crashed_-_locking_protocol/comment_7_cb7ce88ae3d77b9ba0a4e33c2321a3e1._comment [deleted file]
doc/bugs/WSL_adjusted_braches__58___smudge_fails_with_sqlite_thread_crashed_-_locking_protocol/comment_8_cb7ce88ae3d77b9ba0a4e33c2321a3e1._comment [new file with mode: 0644]

diff --git a/doc/bugs/WSL_adjusted_braches__58___smudge_fails_with_sqlite_thread_crashed_-_locking_protocol/comment_7_cb7ce88ae3d77b9ba0a4e33c2321a3e1._comment b/doc/bugs/WSL_adjusted_braches__58___smudge_fails_with_sqlite_thread_crashed_-_locking_protocol/comment_7_cb7ce88ae3d77b9ba0a4e33c2321a3e1._comment
deleted file mode 100644 (file)
index c63d1c2..0000000
+++ /dev/null
@@ -1,15 +0,0 @@
-[[!comment format=mdwn
- username="joey"
- subject="""comment 7"""
- date="2021-10-20T17:04:09Z"
- content="""
-git-annex opened a second connection to the database for writes, in
-addition to the connection it used for reads. That seems likely to be
-involved in whatever locking problem there is on WSL.
-
-Commit [[!commit d0ef8303cf8c4f40a1d17bd134af961fd9917ca4]] eliminates that
-second connection. But there's some chance I'll have to revert it.
-
-If you test, please include `git-annex version` output
-so I can make sure you have a version with that change.
-"""]]
diff --git a/doc/bugs/WSL_adjusted_braches__58___smudge_fails_with_sqlite_thread_crashed_-_locking_protocol/comment_8_cb7ce88ae3d77b9ba0a4e33c2321a3e1._comment b/doc/bugs/WSL_adjusted_braches__58___smudge_fails_with_sqlite_thread_crashed_-_locking_protocol/comment_8_cb7ce88ae3d77b9ba0a4e33c2321a3e1._comment
new file mode 100644 (file)
index 0000000..88a1727
--- /dev/null
@@ -0,0 +1,24 @@
+[[!comment format=mdwn
+ username="joey"
+ subject="""comment 7"""
+ date="2021-10-20T17:04:09Z"
+ content="""
+@asakurareiko oh excellent news!
+
+Before I saw that it's apparently fixed, I noticed that
+git-annex opened a second connection to the database for writes, in
+addition to the connection it used for reads. That seems likely to be
+involved in whatever locking problem there is on WSL. While maybe I already
+fixed the main one, the fact that fix works makes me even more suspicious
+about situations where there are multiple database connections.
+
+Commit [[!commit d0ef8303cf8c4f40a1d17bd134af961fd9917ca4]] eliminates that
+second connection. But there's some chance I'll have to revert it.
+
+If you test, please include `git-annex version` output
+so I can make sure you have a version with that change.
+
+I also wonder if this means that running 2 git-annex commands
+simulantaneously will sometimes result in the same sqlite problem, despite
+these fixes.
+"""]]