[rfc][icedtea-web] immutbale transaltor

[Omair Majid]

* Jiri Vanek <jvanek at redhat.com> [2014-09-17 07:00]:
> Attached patch is changing our new Transaltor to be immutable.

That's great! I love immutable classes!

> And so get rid of enum in favor of class, and so use handler idiom for creation....

Wait, what. How does this follow from the above?

> Personally,  why people use enums for singletons?!?!?!

Please see Effective Java for an explanation. There's a summary on
StackOverflow [0] too.

> Well very simple
> singletons .. yes, but once there is some logic... Ouch they are so so
> *clumsy* ... So clamsy taht they can not be even properly tested ( no
> instance!, no overrload!).

> And once singleton is not immutbale.... Is it still singleton?

A singleton is a class designed to be instantiated (at max) once. It
should not be instantiated again. If you class is, by design, acceptable
to instantiate more than once, then it's not a singleton.

I am not sure how immutability affects singleton. I mean, most
singletons are mutable, but it might be possible that you have a global
shared resources that can be initialized only once and is basically
immutable (or a constant) after.

> +    Translator() {

> +    Translator(String s) {

> +    Translator(ResourceBundle resources) {

Is it intentional that these constructs are now visible package-wide?
Because this kind of violates the singleton instance requirement. I
mean, it's fine if you want to lean more heavily towards having a
testable class rather than
hell-will-break-loose-if-another-instance-of-this-class-is-created
class.

Thanks,
Omair

[0] http://stackoverflow.com/a/71399/3561275
-- 
PGP Key: 66484681 (http://pgp.mit.edu/)
Fingerprint = F072 555B 0A17 3957 4E95  0056 F286 F14F 6648 4681