Yes, I did not mean you shouldn't add declarations, just that you
shouldn't go to any extra effort to do so.
On 2008-03-21, at 10:28 EDT, André Bargull wrote:
(Btw, for your recent change "Redefine LzCursor using 'modern O-O'",
you've used type declarations for the "LzCursorInterface".)
Proposal:
For both LPP-737 and LPP-5648 (trunk), I won't add type
declarations. But I'll mail the typed version of LzParsedPath to
Henry (?) to integrate it into devildog, because there you'll need
it. And after the devildog merge, I'm going to commit my changes for
LPP-654.
So, all I need to do now, is to revert my local version to the
different versions for trunk and devildog. Ok so far?
On 3/21/2008 1:49 PM, P T Withington wrote:
The question is: if we add type declarations in ringding, where
they will not be checked, are we laying a landmine for integrating
to devildog where they will? I tend to think we should not add
type declarations in ringding, for now. We expect to bring
devildog back to trunk "real soon now", at which point, adding type
declarations will be meaningful and testable.
On 2008-03-20, at 20:16 EDT, André Bargull wrote:
I'm going to change LzParsedPath to fix LPP-737 and LPP-5648 (for
trunk), should I add type declarations wherever possible, even if
it isn't the devildog branch?
- André
As of r8198 in trunk, you can now write:
function foo (x:Number, y:String = "", ...rest) { ... }
That is, you can add type declarations (which are ignored at
present for JS1-based run times), you can give default values
for optional arguments, and you can have a "rest" argument that
will pick up any additional arguments into an array.
This is the syntax that will eventually be supported by
Javascript 2, but we (thanks Don!) have implemented it in our
script compiler so that it can be used in Javascript 1 runtimes
too.
There is a test in test/optarg.lzx, if you want to see more
examples.
