Closed
Bug 1097953
Opened 10 years ago
Closed 8 years ago
Combine PluginWidgetChild and PluginWidgetProxy
Categories
(Core Graveyard :: Plug-ins, defect)
Core Graveyard
Plug-ins
Tracking
(e10s-)
RESOLVED
INCOMPLETE
Tracking | Status | |
---|---|---|
e10s | - | --- |
People
(Reporter: jimm, Unassigned)
References
Details
These two have different lifetimes, but billm assures me nsISupports and ipdl can live together harmoniously if ipdl uses ref counting in its management.
This is a non-blocking e10s bug, probably would make a good first bug too.
Comment 1•10 years ago
|
||
I really hate reference-counted IPDL objects, see bug 1096536 for a discussion. My opinion is that it's pretty much always better to have separate refcounted and IPDL objects.
billm do you disagree?
Flags: needinfo?(wmccloskey)
In my opinion, the problem of having IPDL actors whose lifetimes aren't a good match for IPDL won't go away. If we tell people to use separate objects, then we'll just have a different problem where people create two objects that point to each other. When people forget to clear these pointers at the right times, they'll create use-after-free errors.
I think IPDL needs to manage this stuff itself. It seems easier to manage if it's all done in one object, but I'm sure there's a way to do it with two objects too.
Flags: needinfo?(wmccloskey)
Comment 3•10 years ago
|
||
I am much less worried about UAF errors: because IPDL has explicit shutdown machinery with ActorDestroy, there is well-known place to do that cleanup, and the null checks that might be required are more obvious when writing and reviewing code.
Comment 4•8 years ago
|
||
Resolving old bugs which are likely not relevant any more, since NPAPI plugins are deprecated.
Status: NEW → RESOLVED
Closed: 8 years ago
Resolution: --- → INCOMPLETE
Updated•3 years ago
|
Product: Core → Core Graveyard
You need to log in
before you can comment on or make changes to this bug.
Description
•