bzr too slow

[Martin Pool]

On 11 Jan 2006, Denys Duchier <duchier at ps.uni-sb.de> wrote:
> John Arbash Meinel <john at arbash-meinel.com> writes:
> 
> > I'm also concerned about the semantics of finish() versus commit(). Do
> > we always call Transaction.finish() whether we commit() or abort()?
> > Do we need a separate callback queue for things to run if the
> > transaction succeeds versus what happens if it is canceled?
> 
> After thinking about it some more, I believe that there are two kinds of tasks
> that may need to be executed at the end of a transaction:
> 
>     - commit tasks
>     - cleanup tasks (in which I include "cache updates")
> 
> probably "commit tasks" should be FIFO and "cleanup tasks" LIFO.  "commit tasks"
> are executed only if we choose to commit, but "cleanup tasks" are always
> executed.

Right, and you could possibly also have "rollback tasks", which are run
if the transaction is cleaned up without being committed.

Of course you can't strictly rely on all of them running - the process
might segfault or lose its network connection - but putting them in a
systematic framework doesn't seem to make that problem any harder.

-- 
Martin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060112/7d9b7939/attachment.pgp