Austin Clements <aclements@csail.mit.edu> writes:

> On Mon, 03 Dec 2012, david@tethera.net wrote:
>> From: David Bremner <bremner@debian.org>
>>
>> It's a bit annoying to call tar twice, but we cache the results so it
>> isn't as bad as it could be.
> Why not --strip-components=1 and unpack both mail/ and tags/ into a
> single, shared corpus cache directory in one call to tar?  Since you're
> going to cp -lr things anyway, you can structure the corpus cache
> however is convenient.

It's a good suggestion. The only downside is duplicating the tags. I
suppose on the scale of things that isn't a very big waste of space; the
tag corpus is currently 300k, which is dwarfed by even the "small"
corpus.

d