summaryrefslogtreecommitdiff
path: root/doc/todo/parallel_possibilities.mdwn
diff options
context:
space:
mode:
authorGravatar Joey Hess <joey@kitenet.net>2010-10-25 11:47:45 -0400
committerGravatar Joey Hess <joey@kitenet.net>2010-10-25 11:47:45 -0400
commit81f71e57b9ac68b8d79c30fb27a22c5e3941fcee (patch)
tree69b2f391f34a47651955d7ad2506edb35dd30618 /doc/todo/parallel_possibilities.mdwn
parentfed1d3c1f534e01414292cd99724e15f5b19cda7 (diff)
reorg
Diffstat (limited to 'doc/todo/parallel_possibilities.mdwn')
-rw-r--r--doc/todo/parallel_possibilities.mdwn13
1 files changed, 13 insertions, 0 deletions
diff --git a/doc/todo/parallel_possibilities.mdwn b/doc/todo/parallel_possibilities.mdwn
new file mode 100644
index 000000000..178f95021
--- /dev/null
+++ b/doc/todo/parallel_possibilities.mdwn
@@ -0,0 +1,13 @@
+One of my reasons for using haskell was that it provides the possibility of
+some parallell processing. Although since git-annex hits the filesystem
+heavily and mostly runs other git commands, maybe not a whole lot.
+
+Anyway, each git-annex command is broken down into a series of independant
+actions, which has some potential for parallelism.
+
+Probably they would need to be split further. Each action currently has 3
+distinct phases, basically "check", "do", and "record". If the check action
+returned a do action that returned a record action, then it could easily
+make sense to parallelize the check actions and start on the do actions
+(which probably won't parallelize well) while they are still being
+generated, and possibly parallelize the record actions at the end.