On Fri, Aug 29, 2008 at 10:11, Ernst Bunders <[EMAIL PROTECTED]> wrote:
> hello Michiel
>
> I'm not sure that a node should be recreated under these
> circumstances. I think that when this happens, the transaction should
> somehow fail. The behaviour you propose is very contra intuitive, and
> can lead to a lot of confusion with reappearing 'ghost' nodes.
> I think I would like to know it if i am committing a transaction that
> uses nodes that have been deleted.
The information would, in the current situation, be a very horrible
exception, which cannot be recovered from. The exception at the moment
even comes from the database.
It is precisely the same with changing content. This is simply
overwritten by the transaction, whether is was changed by something
else in the mean time or not. In the editwizards you can also opt for
you users to be 'informed' ('Dont panic!'), but I think the general
sentiment is, that that is hardly better.
Btw, I do not think this happens very often so I can't imagine 'a lot
of' confusion. I would not even know how to know that a node is such a
recreated node, unless I happen to have remembered the node-number.
Similar content could be created by other people all the time. It can
be confusing, but well, life can be a bit confusing sometimes.
Anyhow, IMHO a horrible exception if it happens is _really_ confusing,
and worse so, it can make other work, in the same transaction,
disappear, which is downright unacceptable.
So, I could agree that it would be nice if I could _somehow_ know that
if I commit a transaction, that something in it was changed/deleted in
the mean time, but since I have no idea_how_ that then should happen,
I think the solution of simply ignoring the complete fact and
obedientely performing the requested task, is not so bad, I'd rather
have my stuffed saved, not knowing that something odd occured, than
having to restart all over again...
Michiel
--
mihxil' http://meeuw.org
nl_NL eo_XX en_US
_______________________________________________
Developers mailing list
Developers@lists.mmbase.org
http://lists.mmbase.org/mailman/listinfo/developers
