Bug 105592 – Perform cleanup in WebProcess if the NetworkProcess crashes

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKit2

[Brady Eidson]

Perform cleanup in WebProcess if the NetworkProcess crashes.

Any outstanding ResourceLoaders waiting on loads from the network process should probably fail, and for now I'm thinking fail as "cancelled"

In radar as <rdar://problem/12903117>

[Brady Eidson]

Created attachment 180532 [details]
Patch v1

[Alexey Proskuryakov]

Comment on attachment 180532 [details]
Patch v1

Discussed this in person.

1. It should be a real load failure, not cancellation.
2. cancel() should call into NetworkProcess (to actually cancel in normal case), which we don't want to do here.

[Brady Eidson]

(In reply to comment #2)
> (From update of attachment 180532 [details])
> Discussed this in person.
> 
> 1. It should be a real load failure, not cancellation.

Yup.

> 2. cancel() should call into NetworkProcess (to actually cancel in normal case), which we don't want to do here.

Filed <rdar://problem/12924973> for that, will import in to bugzilla later.

[Brady Eidson]

Created attachment 180535 [details]
Patch v2

[WebKit Review Bot]

Attachment 180535 [details] did not pass style-queue:

Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'Source/WebKit2/ChangeLog', u'Source/WebKit..." exit_code: 1
Source/WebKit2/WebProcess/WebCoreSupport/WebErrors.h:30:  Code inside a namespace should not be indented.  [whitespace/indent] [4]
Source/WebKit2/Shared/API/c/WKError.h:46:  enum members should use InterCaps with an initial capital letter.  [readability/enum_casing] [4]
Total errors found: 2 in 9 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Brady Eidson]

http://trac.webkit.org/changeset/138387

[Brady Eidson]

Updated localizable strings in followup http://trac.webkit.org/changeset/138394

[Ryosuke Niwa]

http://trac.webkit.org/changeset/138394
@@ -621,5 +639,5 @@
 
 /* accessibility role description for a file upload button */
-"file upload" = "file upload button";
+"file upload" = "file upload";

caused platform/mac/accessibility/file-upload-button-subrole.html to start failing. Should we just rebaseline or revert the change?

[Brady Eidson]

(In reply to comment #8)
> http://trac.webkit.org/changeset/138394
> @@ -621,5 +639,5 @@
> 
>  /* accessibility role description for a file upload button */
> -"file upload" = "file upload button";
> +"file upload" = "file upload";
> 
> caused platform/mac/accessibility/file-upload-button-subrole.html to start failing. Should we just rebaseline or revert the change?

That happened simply because I ran update-localizable-strings (or whatever it is actually called) which apparently nobody had done in awhile.

If code changed that caused the generated string to be "file upload button" and that code was correct, then we should rebaseline.

I have no idea how to explore that.

[Sam Weinig]

(In reply to comment #9)
> (In reply to comment #8)
> > http://trac.webkit.org/changeset/138394
> > @@ -621,5 +639,5 @@
> > 
> >  /* accessibility role description for a file upload button */
> > -"file upload" = "file upload button";
> > +"file upload" = "file upload";
> > 
> > caused platform/mac/accessibility/file-upload-button-subrole.html to start failing. Should we just rebaseline or revert the change?
> 
> That happened simply because I ran update-localizable-strings (or whatever it is actually called) which apparently nobody had done in awhile.
> 
> If code changed that caused the generated string to be "file upload button" and that code was correct, then we should rebaseline.
> 
> I have no idea how to explore that.

Lets just update the results and check with Chris Fleizach after winter shutdown.

[chris fleizach]

We should rename the key (and string) to be
"file upload button"

(In reply to comment #10)
> (In reply to comment #9)
> > (In reply to comment #8)
> > > http://trac.webkit.org/changeset/138394
> > > @@ -621,5 +639,5 @@
> > > 
> > >  /* accessibility role description for a file upload button */
> > > -"file upload" = "file upload button";
> > > +"file upload" = "file upload";
> > > 
> > > caused platform/mac/accessibility/file-upload-button-subrole.html to start failing. Should we just rebaseline or revert the change?
> > 
> > That happened simply because I ran update-localizable-strings (or whatever it is actually called) which apparently nobody had done in awhile.
> > 
> > If code changed that caused the generated string to be "file upload button" and that code was correct, then we should rebaseline.
> > 
> > I have no idea how to explore that.
> 
> Lets just update the results and check with Chris Fleizach after winter shutdown.