making read locks noops

[Martin Pool]

On  1 Mar 2006, Robert Collins <robertc at robertcollins.net> wrote:
> So whats needed to make read locks noops ? and where is it safe?

I've done that in one branch (which, hm?) and it at least passes the
tests - but then they don't exercise concurrency as much as they could,
and even then it's a very hard thing to test exhaustively.

> WorkingTree seems less safe because while we never lock user files, a
> changed inventory will give us different paths to read user files from,
> so we would like the current inventory to be fixed during a read
> transaction.... but things like push might be altered to not lock the
> working tree at all.

I think we should hold the inventory for the duration of the
transaction.  Obviously we can't control modifications by other programs
of the working directory so we just need to try to be robust against it.

-- 
Martin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: Digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060302/ead1a351/attachment.pgp