Comments on Tapestry Central: Duh! Annotation values as constants, not literals

Well, specifically: T5 IoC will not register this...

2007-04-26T13:58:00.000-07:00

Well, specifically: T5 IoC will not register this as an error until it realizes your service, at which point it will see the error, and identify the correct possible values.

T5 IoC has also tested to death this issue, which is why I would much rather lean on that code that endlessly write ThreadLocal-based code to accomplish the same thing. Less code == good (if you can trust it).

whereas @Scope("prthread") would not be caught unt...

2007-04-26T13:47:00.000-07:00

whereas @Scope("prthread") would not be caught until runtime.

...if then. One of the major problems of the run-time validation which seems to be so popular is that it's REALLY tricky to tell if you've screwed something up. You have to make sure you have a unit test which hits the particular scenario and successfully fails in the case of your particular typo.

Worse, these kind of unit tests are notoriously heavyweight to write. For instance, in this case you'd have to have a pretty fancy unit test which involved firing off two different threads and interrogating them to find the difference: that's a lot of work. And if a unit test is a lot of work to write, it's all the more likely to be abandoned -- right when you needed it most!

Strangely, that part I knew.

2007-04-24T12:45:00.000-07:00

Strangely, that part I knew.

you can use Enums too.... seen in JPA g,kris

2007-04-24T12:43:00.000-07:00

you can use Enums too.... seen in JPA

g,
kris