On 2011-08-22, at 1:38 PM, Scott Sauyet wrote:
> Laurie Harper wrote:
>> If you start using 'internal' identifiers in your events, your
>> sandbox is leaking. Why go to the trouble of hiding the object
>> types and decoupling everything if you're just going to couple
>> it up again by exposing implementation details in string
>> identifiers?
>
> I think it's simply a matter of namespacing shared keys in a publish-
> subscribe registry. It's not a matter of exposing implementation
> details.
There's a difference between an event descriptor that says 'secret code
entered', one that says 'secret code was typed in using a keyboard', and one
that says 'entry of secret code was detected by the "keyboard" module'...
In other words, the event naming should be based on the semantics of what
happened, not on what module noticed it happen.
> Of course your modules should be decoupled as much as is reasonably
> possible, but a namespacing scheme like Jarek described ("keyboard-
>> onSecretCodeEntered") seems reasonable. You don't have to know
> anything about the keyboard module to listen for such events. For all
> you know there is no module named "keyboard", but, through the
> sandbox, you can listen whenever something generates this particular
> event. The namespacing is not central, but it makes it easier. For
> example, in an application that represents some card game, it's really
> nice to be able to distinguish "deck->draw" events from "screen->draw"
> ones.
Those are both examples of semantic event descriptors; I agree that namespacing
your events makes sense. You just need to be sure you're using something that
describes what happened, not how it happened or where it was detected.
Tying event names to the name of the module generating them makes it impossible
to switch modules out, thus defeating the objective of decoupling modules.
Being over-specific with event names also reduces your flexibility; what if you
want to add support for entering a secret code via a bio-metric device? The
word 'keyboard' in the event name is suddenly kinda dubious... Use a
namespacing scheme that reflects *just* the semantics you want to capture.
L.
>
> -- Scott
>
> --
> To view archived discussions from the original JSMentors Mailman list:
> http://www.mail-archive.com/[email protected]/
>
> To search via a non-Google archive, visit here:
> http://www.mail-archive.com/[email protected]/
>
> To unsubscribe from this group, send email to
> [email protected]
--
Laurie Harper
http://laurie.holoweb.net/
--
To view archived discussions from the original JSMentors Mailman list:
http://www.mail-archive.com/[email protected]/
To search via a non-Google archive, visit here:
http://www.mail-archive.com/[email protected]/
To unsubscribe from this group, send email to
[email protected]
