multiple machine tagging

Jesse Rosenthal jrosenthal at jhu.edu
Thu May 20 09:28:05 PDT 2010


On Thu, 20 May 2010 09:25:42 +0100, David Edmondson <dme at dme.org> wrote:
> What's the current state of the art in merging tags from multiple
> machines?

About the same, I think. I had started trying to figure out a way to do a
smarter dump (only dumping changed info) for my distributed
tagging/bug-tracking idea, and had some early success, though the python
implementation kills some of the speed gains. (Do play with it though!)
But in your case, dumping probably isn't the lag so much as restoring,
right?

What about: 
nm dump master>master-dump; nm dump copy >copy-dump; 
diff copy-dump master-dump | grep "^>" | etc...

This could all be scripted with scp or curl to do the networking
between machines.

Smarter (history- and namespace-aware) dumping could only improve this,
of course, and is very possible. But the abovr approach seems like it
would remove the restore bottleneck (unless I misunderstand how restore
works with unlisted messages).

Best,
Jesse


More information about the notmuch mailing list