Just found this task: "http://www.openlaszlo.org/jira/browse/LPP-2679";.
On 12/11/2008 12:55 AM, André Bargull wrote:
On 12/11/2008 12:41 AM, Henry Minsky wrote:
It turns out that's ok for fetching data for display, but not
ok for transactional behaviors. That's why we added
"queuerequests", which tried to offer that transaction-ality.
As such, queuerequests was supposed to:
* guarantee that a request was made for each request call on
the dataset
* guarantee that the program would be notified in the correct
order of the responses, regardless of the network latencies
I remember writing the test for that, though god knows what
happened to it :)
A
Hmm, ok, I'll see if I can find excavate any old tests. It sounds like
the responses
from the server have to be queued up in the client, in case they
arrived out of order.
All queuing happens for swf in LzLoader, see "LzLoader#expectedList".
If we want to support this feature (again?), we should move that code
out of LzLoader into a runtime independent class.
