Michael Hunger wrote:
> why exposing the internal structure of the Message that way? Wouldn't it be
> better to have value composites for the
> header(s) which in turn could have value composites for their fields. the
> same for attachments. Otherwise I'd prefer a
> real collection type (Header) instead of having a map there.
Sounds like a great idea. I'll fix it.
> That brings me to the question. Why arent there keyed (role supporting)
> associations in qi4j? So that I can have a
>
> Person father = ...
> Person mother = ...
> Person child = ...
> child.relatives().add("father",father)
> child.relatives().add("mother",mother)
We considered this initially, but skipped it for now. If we can figure
out how to do it well we can add it later I think. Any suggestions are
welcome.
> I don't know if the message store should send them automatically? Perhaps I'd
> only like store a draft or a copy.
Then you wouldn't put them in the "Outbox", but in a "Draft" folder.
Messages should only be sent automatically if they are in an "Outbox".
We'll use the same thinking as in email clients is my suggestion.
> I'd
> also use the same message store (inbox) for the messages received.
Depends on what you mean by "message store". If we separate between
"message store" and the actual transports (SMTP, POP3, XMPP, etc.) then
I agree, since messages in one message store could be sent using several
transports.
> Should there be a state property - draft, sending, sent, failed) property.
Nope. This should be handled by using Folders, for all of the above.
Much more flexible and easier to then add rules to those folders (e.g.
"Failed sending Message so put it in Failed folder which triggers
'Notify admin' XMPP message").
Makes sense?
/Rickard
_______________________________________________
qi4j-dev mailing list
[email protected]
http://lists.ops4j.org/mailman/listinfo/qi4j-dev
