351514 - Finalize yield syntax to match ES4/JS2 proposal

Status: VERIFIED FIXED

(Core :: JavaScript Engine, defect, P1)

Description

[Brendan Eich [:brendan]]

This tracks Python 2.5.  Currently we allow

  function f() { yield g(yield 1, 2) }

but the inner yield must be parenthesized, per PEP 342, Python 2.5, and the ES4 / JS2 proposal.  Trivial patch next.

/be

Comment 1

[Brendan Eich [:brendan]]

Attachment: fix

Comment 2

[Brendan Eich [:brendan]]

This is important to get right in js1.7/mozilla1.8.1/firefox2 -- we don't want to be hobbled with a compatibility requirement down the road based on js1.7 supported unparenthesized yield expressions as actual arguments.

/be

Comment 3

[Brendan Eich [:brendan]]

Comment on attachment 236932 [details] [diff] [review]
fix

This is safe, it merely adds a more specific error, registerizes PN_LAST(pn) in pn2 in the existing bad-yield-syntax case, and adds identical error test-and-report code to the ArgumentList parser.

/be

Comment 4

[Brendan Eich [:brendan]]

Fixed on trunk.

/be

Comment 5

[Brendan Eich [:brendan]]

Attachment: fix for 1.8 branch

Comment 6

[Bob Clary [:bc] (inactive)]

Checking in regress-351514.js;
/cvsroot/mozilla/js/tests/js1_7/geniter/regress-351514.js,v  <--  regress-351514.js
initial revision: 1.1

Comment 7

[Mike Beltzner [:beltzner, not reading bugmail]]

Comment on attachment 237043 [details] [diff] [review]
fix for 1.8 branch

a=beltzner on behalf of 181drivers

Comment 8

[Brendan Eich [:brendan]]

Fixed on the 1.8 branch.

/be

Comment 9

[Bob Clary [:bc] (inactive)]

verified fixed 1.9a1_2006090707 windows/mac*/linux

Comment 10

[Bob Clary [:bc] (inactive)]

verified fixed 1.8 1.9 20060909 windows/mac*/linux

Comment 11

[Brendan Eich [:brendan]]

Attachment: fix js1_7/geniter/regress-345855.js

Comment 12

[Bob Clary [:bc] (inactive)]

Comment on attachment 238592 [details] [diff] [review]
fix js1_7/geniter/regress-345855.js

print will try to print to a real printer if this is run in the browser. writeLineToLog() is preferred in non-e4x tests although we _could_ do what e4x does and redefine print.

These new values are the correct, non-error producing values? Would the old values cause errors?

Comment 13

[Brendan Eich [:brendan]]

Sorry, forgot I added those prints -- it was impossible to tell what was failing before I did that.  Maybe the actual and expected values could encode something about the specific test?  Anyway, just passing through here.

The upshot is that except when used as a statement, and when used on the right hand side of assignment, yield expressions must be parenthesized.  Previously accepted under-parenthesized yield (specifically as a single actual parameter to a function, foo(yield bar) e.g.) is not allowed now.

/be

Comment 14

[Bob Clary [:bc] (inactive)]

Attachment: fixier js1_7/geniter/regress-345855.js

The numeric labels were meant to indicate which section failed but I went a head and changed them. This also adds the syntax errors.

Comment 15

[Brendan Eich [:brendan]]

Comment on attachment 238595 [details] [diff] [review]
fixier js1_7/geniter/regress-345855.js

Thanks, works for me.

/be

Comment 16

[Bob Clary [:bc] (inactive)]

Checking in regress-345855.js;
/cvsroot/mozilla/js/tests/js1_7/geniter/regress-345855.js,v  <--  regress-345855.js
new revision: 1.2; previous revision: 1.1
done