Thanks for the reviews!
I've uploaded a new webrev with all the changes you suggested. For
Type.typeFor I added a call to a new Context.isStructureClass method.
Note that structure classes are direct subclasses of ScriptObject, so we
use Type.SCRIPT_OBJECT in that case.
http://cr.openjdk.java.net/~hannesw/8135190/webrev.01/
Hannes
Am 2015-09-18 um 15:09 schrieb Sundararajan Athijegannathan:
* On naming: how about getHigh and getLow in Splittable?
* Types.typeFor can fallback to name pattern (JO subclass) and use
JO.class (when getting CNFE) or context's sharedloader to do
Class.forName
Other than that +1
-Sundar
On 9/18/2015 6:33 PM, Attila Szegedi wrote:
Few minor remarks:
* AssignSymbols.isSplitLiteral uses !(splitRanges == null ||
splitRanges.isEmpty()) while everywhere else in the code the check is
(ranges != null).
* loadSplitLiteral: seems like there's a stray comment: //
method.load(literalType, literalSlot);
* Terminology: Splittable is described as a "marker interface” yet it
has a method.
Other than these small things, +1.
Attila.
On Sep 18, 2015, at 12:03 PM, Hannes Wallnoefer
<hannes.wallnoe...@oracle.com> wrote:
Please review JDK-8135190: Method code too large in Babel browser.js
script:
http://cr.openjdk.java.net/~hannesw/8135190/
This makes object literals splittable. I tried to share as much code
as possible with array literal splitting so I had to introduce a
couple of new interfaces.
Hannes
