Bug 16889
| Summary: | REGRESSION (r29425): Canvas-based graphing calculator fails to run | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Oliver Hunt <oliver> |
| Component: | JavaScriptCore | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED DUPLICATE | ||
| Severity: | Normal | CC: | ddkilzer, eric, ggaren, mjs, oliver, zwarich |
| Priority: | P1 | Keywords: | InRadar, NeedsReduction, Regression |
| Version: | 528+ (Nightly build) | ||
| Hardware: | All | ||
| OS: | All | ||
| URL: | http://www.calc5.com/ | ||
| Bug Depends on: | 17164 | ||
| Bug Blocks: | |||
Oliver Hunt
The graphing calcualtor at calc5.com now completely fails to load or run, it appears to produce "undefined value" exception when it did not used to.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Oliver Hunt
Regression occurs between r29414 and r29438
Oliver Hunt
Erk, this includes the activation patch, which could easily be the cause of the badness :(
Cameron Zwarich (cpst)
I can confirm that it doesn't work. It also doesn't work with my latest patch for bug 16868. Once that gets landed, I'll fix this. It doesn't look like it'll be too bad. The JS isn't obfuscated in any way like the Google stuff.
Oliver Hunt
yay google code! /me curses obfuscation tools
Cameron Zwarich (cpst)
I can confirm our suspicions that the bug is caused by r29425 (the ActivationImp tear-off patch). I patched a clean r29414 tree with it and it no longer worked.
Cameron Zwarich (cpst)
This is not fixed by Maciej's fix for bug 16909.
Oliver Hunt
That makes me sad :(
David Kilzer (:ddkilzer)
<rdar://problem/5696255>
David Kilzer (:ddkilzer)
(In reply to comment #5)
> I can confirm our suspicions that the bug is caused by r29425 (the
> ActivationImp tear-off patch). I patched a clean r29414 tree with it and it no
> longer worked.
The ActivationImp issue is now resolved, but the results for the calculator don't display properly (unlike on Firefox). I think this bug should be kept open, and have its focus change to resolve the issue with displaying results.
Oliver Hunt
Dave, what makes you think that the activation imp issue is resolved? afaict the activation patch introduced the undefined variable exception, and that is still present in ToT -- the various activation related fixes were to stop crashes in GC, not fix correctness bugs -- none even touched correctness
Cameron Zwarich (cpst)
(In reply to comment #9)
> (In reply to comment #5)
> > I can confirm our suspicions that the bug is caused by r29425 (the
> > ActivationImp tear-off patch). I patched a clean r29414 tree with it and it no
> > longer worked.
>
> The ActivationImp issue is now resolved, but the results for the calculator
> don't display properly (unlike on Firefox). I think this bug should be kept
> open, and have its focus change to resolve the issue with displaying results.
Like Oliver said, the GC marking issues were resolved, but this bug didn't change one bit. It is such a random error that it must be caused by some strange behaviour introduced by the ActivationImp tear-off. There are a few of these bugs now, but my previous attempts to reduce them failed. I will try again with the calculator now that I have more free time because I really want to see this resolved.
Cameron Zwarich (cpst)
This bug is fixed by the patch for bug 17164.
Oliver Hunt
*** This bug has been marked as a duplicate of 17164 ***