Bug 43372 - [wx] Crash during HitTest call
Summary: [wx] Crash during HitTest call
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: WebKit wx (show other bugs)
Version: 528+ (Nightly build)
Hardware: All All
: P2 Normal
Assignee: Nobody
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-08-02 13:20 PDT by Malcolm MacLeod
Modified: 2010-08-03 11:46 PDT (History)
1 user (show)

See Also:


Attachments
Patch to fix initialisation crash (1.09 KB, text/plain)
2010-08-02 13:31 PDT, Malcolm MacLeod
kevino: review+
kevino: commit-queue-
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Malcolm MacLeod 2010-08-02 13:20:34 PDT
Calling HitTest on a wxWebView results in a crash due to a) destruction of an uninitialized variable and b) (after a is fixed) deletion of a variable that has already been freed, presumably due to some kind of smart pointer.
This patch fixes both of these issues so that HitTest can be used ina  functional manner again.
Comment 1 Malcolm MacLeod 2010-08-02 13:31:08 PDT
Created attachment 63249 [details]
Patch to fix initialisation crash
Comment 2 Alexey Proskuryakov 2010-08-03 07:34:32 PDT
Comment on attachment 63249 [details]
Patch to fix initialisation crash

Would you be willing to submit a patch for review, as described in <http://webkit.org/coding/contributing.html>?

Please don't comment out code - just delete it if it's wrong.
Comment 3 Kevin Ollivier 2010-08-03 11:44:58 PDT
Thanks! Landed in r64568. 

A couple remarks for future patches. One, there is a 'patch' checkbox which lets the system know the attachment is a patch, and setting ? on the review queue marks it as needing review. (It also actually initiates try-bots which attempt to compile the patch.)

Also, typically patches are supposed to contain a ChangeLog entry, which can be auto-generated using WebKitTools/Scripts/prepare-ChangeLog. Typically patches missing a ChangeLog entry are supposed to be given a bad review and sent back for reworking, but as this was fixing a crash, I felt it was best to be expedient and land this ASAP, adding the ChangeLog entry myself.
Comment 4 Kevin Ollivier 2010-08-03 11:46:47 PDT
(In reply to comment #3)
> Thanks! Landed in r64568. 
> 
> A couple remarks for future patches. One, there is a 'patch' checkbox which lets the system know the attachment is a patch, and setting ? on the review queue marks it as needing review. (It also actually initiates try-bots which attempt to compile the patch.)
> 
> Also, typically patches are supposed to contain a ChangeLog entry, which can be auto-generated using WebKitTools/Scripts/prepare-ChangeLog. Typically patches missing a ChangeLog entry are supposed to be given a bad review and sent back for reworking, but as this was fixing a crash, I felt it was best to be expedient and land this ASAP, adding the ChangeLog entry myself.

Err, bad review is a poor choice of words. :) I mean an r-, which means it isn't ready to land.