WebKit Bugzilla
New
Browse
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED INVALID
44430
[Qt] WebCore crashes from the web process.
https://bugs.webkit.org/show_bug.cgi?id=44430
Summary
[Qt] WebCore crashes from the web process.
Balazs Kelemen
Reported
2010-08-23 09:04:34 PDT
In situations where we want to cast a client class to it's qt version (for example a FrameLoaderClient to a FrameLoaderClientQt) are crashing in the web process of webkit2 since the type of the clients are differs. Currently I do not see a better fix for this then doing some hand written rtti (because we should not rely on rtti support) and check the type of those clients. In the future we should eliminate the layering violations and/or introducing better abstractions to deal with the duality of the API layers but for now the webkit2 project needs a working implementation in the trunk.
Attachments
proposed patch
(9.69 KB, patch)
2010-08-24 01:44 PDT
,
Balazs Kelemen
kenneth
: review-
kenneth
: commit-queue-
Details
Formatted Diff
Diff
View All
Add attachment
proposed patch, testcase, etc.
Kenneth Rohde Christiansen
Comment 1
2010-08-24 01:09:08 PDT
We should just fix the layering violations instead and never do casts like these :)
Balazs Kelemen
Comment 2
2010-08-24 01:44:35 PDT
Created
attachment 65236
[details]
proposed patch
Kenneth Rohde Christiansen
Comment 3
2010-08-24 03:30:26 PDT
Comment on
attachment 65236
[details]
proposed patch First of all, each of these should be separate patches! Secondly, there is another patch from Jesus and Diego fixing some of these issues, but it is not 100% ready yet. Thirdly, this is too hacky and it is better to spend the time on fixing this properly by removing the layering violations or helping Jesus and Diego finish their patch.
Balazs Kelemen
Comment 4
2010-08-24 04:42:05 PDT
(In reply to
comment #3
)
> (From update of
attachment 65236
[details]
) > First of all, each of these should be separate patches! > > Secondly, there is another patch from Jesus and Diego fixing some of these issues, but it is not 100% ready yet.
That patch does not covers the PluginData problem. Do you think we should introduce a PluginDataContext there?
> > Thirdly, this is too hacky and it is better to spend the time on fixing this properly by removing the layering violations or helping Jesus and Diego finish their patch.
I can degrease the hackiness via using dynamic_cast but as I know rtti should be avoided in WebKit (and also in Qt maybe). You should take in account that even if we could solve the layering violations (that is materialized now in the form of casting and abstract client class to the concrete qt one), we won't have the fancy API-s of WebKit in WebKit2 so we should provide a default behavior. I would ask you to help me by discussing what design we should follow to solve these problems. Creating *Context abstractions? Using PlatformStrategies somehow? Merging API of WebKit and WebKit2?
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug