You can clone with
HTTPS or Subversion.
Issue #1388 regressed the following code using globalscript: true:
var a = 'toto';
a = a + 'plop';
I think "globalstrict" is not supposed to force using it in global, but only to allow it.
I made #1452 which contains:
I don't exactly know how to fix issue 1388 though... @antonkovalyov if you have an idea I can try to do it.
I thought of calling block for the global scope too, with an additional boolean argument isGlobal. Do you think it's a good idea? Also we could convert block into using an option object instead of 5 or 6 boolean parameters.
Added some code to make the existing test pass, but as my modified test shows, it's not enough ;)
Trying to find the correct place to put this code... ideally, it would be in statements with a way to not test this if the current statement is a function. (not sure this sentence is easy to read ;) )
Just pushed a new version of the pull request #1452 that fixes both the regression and issue #1388. I don't find it very clean so I'm very open to any change you think would be necessary.
Issue #1451: Fix a regression caused by #1388
Signed-off-by: Anton Kovalyov <firstname.lastname@example.org>