10 Nov
2010
10 Nov
'10
1:24 p.m.
On 11/10/2010 11:08 AM, Vincent Massol wrote:
Why only Document (Event can be about anything)?
This is the name of the existing method. Nobody is proposing this name.
If you don't like it, we can change it in the future.
On Nov 10, 2010, at 10:12 AM, Stefan abageru wrote:
Yes..right now the implementation does a test of instanceOf on the event
received (there is an awful if then else) and selects the type of event
from the class ActivityEventType static fields. By this modification we
can remove that if then else and use classname instead of the strings
from ActivityEventType.
Why do you need to do an instanceof?
The identifier field..I see it as a way so identify and specify more
data about the event itself. For example, i think it is important do
differentiate 2 CommentAddedEvent and not jusrt specify that they
occurred in a list of events and i would do this by this field.
The advantage of this field would be that it would be known in a generic
way and access to it would be easily done. There would be no need of the
sync you said between the listener and the emmiter. I aslways know i
will receive a class extending AbstractFilterableEvent, I can get the
identifier field. Maybe move it to Event itself ?
How do you ensure unicity? And I don't think you're using this field in the
correct way. You seem to be needing extra data about the event in order to perform some
logging/display. The field to use for this is the data field. Having a data field in Event
rather than in EventListener.onEvent() doesn't change anything for the sync... :)
Right, except that it's not backwards compatible with the current
implementation. We used to send the XWikiContext as the "data", and
changing this would mean a very backwards incompatible change, although
a change for the better.
On 11/10/2010 10:57 AM, Vincent Massol wrote:
There are some events that are internal events and I don't think they should be
displayed.
+ I don't know how the display will scale for view Events for ex (which can be
received about 1000 times per second).
Hi Stefan,
On Nov 10, 2010, at 9:08 AM, Stefan abageru wrote:
There would ne no filtering since IMO ActivityStream records the
activities happening in the wiki. Should there be some filtering ? Hi devs !
I thought about adding some changes in Events and Activity Stream.
First of all, I want to remove all the listeners from Activity Stream
and make this class listen to AllEvent, making it capable of cathcing
all events that occur in the wiki (and this way, when adding a new
event, you wouldn't need any modifications to this class in order to
catch it).
How do you filter events to process? If you don't how would you
display for example view events or internal events such as component manager descriptor
added/removed event, or the new event being added by Anca for the XAR import?
Secondly, I would like to modify onEvent function in
the class so that
it uses the simpleClassName of the event occurred in order to log the
event (the eventType would be this simpleClassName...for example
DocumentSavedEvent, AnnotationAddedEvent and so on).
I don't agree with this,
or I don't understand it. You get the Event object so you can easily get the simple
class name by calling event.getClass().getSimpleClassName().
And thirdly i thought about adding one field in
AbstractFilterableEvent
called /identifier/ and make all events extend directly this class.
Again I
don't agree. A notion of event identifier is not linked to a Filterable event IMO but
to the Event class itself.
And
that /identifier/ would be used in a different way by the events
themselves. But it would be used to retrieve infos about that events.
And so, when logging an event, we could use classname + identifier.
For example
AnnotationAddedEvent + "This is my annotation" .
CommentAddedEvent + "This is my comment" or the number of the comment.
I
don't quite agree. If it's supposed to be an identifier it must be unique and
leaving the unicity to the user will not guarantee it's unique. I'm not sure what
you want to do:
1) Be able to differentiate 2 events. For what reason?
2) Be able to have some description of the events. Again for what reason?
Now if you need extra data, this is already taken into
account in the onEvent method:
void onEvent(Event event, Object source, Object data);
The data parameter is there to pass extra data.
The way to use it is to create compound objects if you need to pass several data.
However for easiness of use, I'd agree to transform the signature into:
void onEvent(Event event, Object source, Object... data);
All that said, we have a small issue in that the listener of events need to know what
type of data get passed and thus emitter and listeners must be in sync. Thus a generic
listener will always have a problem since it won't know the type of objects it
receives (since they're different for each event). Thus for me, when you have a
generic listener you also need an EventActionHandler that needs to be registered against
the generic listener so that it know what to do with the extra data passed.
How would you handle anonymous events in the Activity Stream (AS) listener? BTW what
actions do you need to do with received events in the AS?
About anonymous events. I
do not know exactly what you mean by that.
In AS i need to record them by doing a call to
addDocumentActivityEvent(streamName, currentDoc, eventType, msgPrefix +
eventType, params, context); In general I don't like this proposal as it is
because it doesn't address the real problem. It's moving it from onEvent() to a
field in Event, adding even more complexity since there would be two ways of passing
information. I'm -1 on it right now, especially since it breaks API without solving
the issue. We need to address the sync issue.
Thanks
-Vincent
> eventType would the the simple class name and the identifier would go in
> the params list.
>
> Thanks,
> Stefan
>> Thanks
>> -Vincent
--
Sergiu Dumitriu
http://purl.org/net/sergiu/