[elephant-devel] Sqlite-3 backend, threads and transactions

[Ignas Mikalajunas]

On 4/19/06, Robert L. Read <read at robertlread.net> wrote:
>
>     I am a little surprised by your use of clsql:start-transaction; the
> normal idiom for
> this is:
>
>        (with-transaction (:store-controller *store-controller*)
>          (loop for obj in objs
>                for key in keys
>                do (setf (get-value key bt) obj))))
>   (this is an example for testcollections.lisp)
>
> Are you saying in your email that you could not use this way of doing
> things?  The
> with-transaction macro should be indpendent of the type of the
> store-controller.
>

And indeed it is independent, but calls to ele:start-transaction,
ele:commit-transaction and ele:abort-transaction are throwing an error for
me and their code seems to be BDB specific. I would like to use
with-transacton, yet it aborts the transaction on "any non-local exit",
while i want to rollback only when some error occurs. TBNL redirect code
does (throw 'tbnl:tbnl-handler-done), aborting transaction on each redirect.
As in web applications (save-data-to-database)
(redirect-to-avoid-double-post) is a very common pattern, i was getting my
data reverted every time. So i am doing manual commits/rollbacks, depending
on whether the request processing code raised an error or not.

You wrote:
> As sqlite requires each thread to have it's own connection by throwing
> an error if one thread tries to reuse a connection created by another
> thread at least on Ubuntu Dapper. Google says that it depends on a
> compile time flag for libsqlite3.
>
> The current Elephant code has no notion of this, and I'm pretty sure it
> will hand the same connection
> to several threads.  Obviously, you could solve this problem by using
> mutexes to make sure that
> only one thread accesses the controller, but that might be quite
> unpleasant for you code.
>

At the moment i am doing exactly this - making sure that only one thread is
working with elephant at the same time, though it is more of a workaround
than a solution.

If these comments can help you provide anymore information about what your
> possible solutions
> are, please reply so I can think about it before I begin debugging it
> tonight or tomorrow.
>

You can find more info about the issue with sqlite here:

    http://projects.edgewall.com/trac/ticket/2196

hope that helps.

--
Ignas Mikalajūnas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/elephant-devel/attachments/20060420/d1bd5d84/attachment.html>