Issue 14621, 14624, 14625 - Fix bugs in array operator overloading semantics #4686
Conversation
| @@ -64,6 +66,8 @@ Expression *resolveAliasThis(Scope *sc, Expression *e) | |||
| e = e->semantic(sc); | |||
| } | |||
| e = resolveProperties(sc, e); | |||
|
|
|||
| e = (gag && global.endGagging(olderrors)) ? NULL : e; | |||
There was a problem hiding this comment.
if (gag && global.endGagging(olderrors))
e = NULL;would be nicer IMO.
9rnsr
force-pushed
the
fix14621
branch
2 times, most recently
from
851037d
to
99760f3
Compare
|
To @WalterBright and other committers: Please review this to fix regression Issue 14621, toward 2.068-beta2 release. |
|
I really don't like the gag technique, even though I came up with it, it's been nothing but trouble. There's got to be a better way, and there is - the TOKerror method. Also, at this point, this looks like a very large and disruptive change for this late in the beta cycle, especially since this regression was introduced well over a year ago. |
|
Did you decide not to fix issue 14621 in 2.068 release? |
…ng fallback Points: - Shorten the length of problematic gagged semantic path. - Improve ArrayExp as the entry point of array operator overloading. - Reduce stack size and the number of heap-allocated AST nodes for recursive semantic analysis. Also fixes unlisted wrong-code issue. See test20x() in runnable/opover2.d
…bal.errors' on line 752 in file 'statement.c'
|
Right, I think this fix is risky to insert into the 2.068 beta, not worth it because it's an old issue and hasn't caused much problems. |
|
Ok, make sense. |
|
If it's an old issue, please downgrade the bug from regression to something else. |
|
It's not so old. The gagging issue is since 2.066, that introduced when the new operator overloading feature is added. |
|
This PR has been moved to #4948, for the transplanting to stable branch. |
Issue 14621 - [REG2.066] ICE: Assertion failure: 'global.gaggedErrors || global.errors' on line 752 in file 'statement.c'
Issue 14624 - The array operator overloading fallback is not correct
Issue 14625 - opIndex() doesn't work on foreach container iteration
Shorten the length of gagged semantic path, and improve
ArrayExpas the entry point of array operator overloading.