fix #31 in resource-loader #32
Conversation
|
Oh btw do we even need the |
| @Shadow | ||
| protected abstract void load(List<?> resources); | ||
|
|
||
| @Inject( |
There was a problem hiding this comment.
It seems to me that this changes more than is needed, as you could keep the previous implementation and wrap the original.call(...)s in try-catch blocks with the same effect?
There was a problem hiding this comment.
I thought that too at first but WrapOperation evaluates the arguments of the function call in order to capture them, and trying to evaluate the list argument throws an exception, so the mixin is not called at all after that.
There was a problem hiding this comment.
aaaahhh gotcha, alright then, this change looks good
I suppose not. Ideally we'd allow uppercase language codes in those versions. Something for later. |
|
|
||
| @Final | ||
| @Shadow | ||
| private Map<String, String> translations; | ||
|
|
||
| @WrapOperation( | ||
| @Shadow | ||
| protected abstract void load(List<?> resources); |
There was a problem hiding this comment.
nit: please parameterize this properly (should be List<Resources>
Instead of wrapping the
loadcall, the alternate lang formats are injected before the argument callinggetResourcesis evaluated. Any exception thrown from this argument evaluation will no longer prevent loading of the alternate files.This changes the evaluation order of the variants but users of these different file name options (
lang/en_US.lang,lang/en_US.json,lang/en_us.lang,lang/en_us.json) should not expect a specific ordering anyway.