summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorGravatar Joey Hess <joey@kitenet.net>2014-11-03 22:10:56 -0400
committerGravatar Joey Hess <joey@kitenet.net>2014-11-03 22:10:56 -0400
commit99ee9087b77b0c4f2a2d69196a85e91ea2994d73 (patch)
tree012c398d89f38dcf69066449e274c6d4ceb62848
parent15ca60d9280514b3e7960e20a20fb337580e9113 (diff)
devblog
-rw-r--r--doc/devblog/day_230__S3_multipart_round_2.mdwn9
1 files changed, 9 insertions, 0 deletions
diff --git a/doc/devblog/day_230__S3_multipart_round_2.mdwn b/doc/devblog/day_230__S3_multipart_round_2.mdwn
new file mode 100644
index 000000000..a73579df8
--- /dev/null
+++ b/doc/devblog/day_230__S3_multipart_round_2.mdwn
@@ -0,0 +1,9 @@
+More work on S3 multipart uploads, since the aws library got fixed today
+to return the ETAGs for the parts. I got multipart uploads fully working,
+including progress display.
+
+The code takes care to stream each part in from the file and out the
+socket, so I'd hoped it would have good memory behavior. However, for
+reasons I have not tracked down, something in the aws library is causing
+each part to be buffered in memory. This is a problem, since I want to
+use 1 gb as the default part size.