summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorGravatar Joey Hess <joeyh@joeyh.name>2016-02-25 17:12:41 -0400
committerGravatar Joey Hess <joeyh@joeyh.name>2016-02-25 17:12:41 -0400
commit0e4d5208fd02996414cef04a4976aebe1e3785fc (patch)
tree183bb58681c169766dd10e8235f22803dbb1cae4
parenta650aa8b43bf21388c2b5e8ebd922a52801eebd4 (diff)
devblog
-rw-r--r--doc/devblog/day_367__adjusted_branches_proof_of_concept.mdwn18
1 files changed, 18 insertions, 0 deletions
diff --git a/doc/devblog/day_367__adjusted_branches_proof_of_concept.mdwn b/doc/devblog/day_367__adjusted_branches_proof_of_concept.mdwn
new file mode 100644
index 000000000..80576005a
--- /dev/null
+++ b/doc/devblog/day_367__adjusted_branches_proof_of_concept.mdwn
@@ -0,0 +1,18 @@
+Now I have a proof of concept [[design/adjusted_branches]]
+implementation, that creates a branch where all locked files
+are adjusted to be unlocked. It works!
+
+Building the adjusted branch is pretty fast; around 2 thousand files
+per second. And, I have a trick in my back pocket that could double that
+speed. It's important this be quite fast, because it'll be done often.
+
+Checking out the adjusted branch can be bit slow though, since git runs
+`git annex smudge` once per unlocked file. So that might need to be
+optimised somehow. On the other hand, this should be done only rarely.
+
+I like that it generates reproducible git commits so the same adjustments
+of the same branch will always have the same sha, no matter when and where
+it's done. Implementing that involved parsing git commit objects.
+
+Next step will be merging pulled changes into the adjusted branch, while
+maintaining the desired adjustments.