[RFC][plugin]: fix concurrency problems in PAV.java

[Deepak Bhole]

* Denis Lila <dlila at redhat.com> [2011-04-14 10:07]:
> > NetxPanel inherits from AVP, so why is the cast needed? Does NetxPanel
> > override the method?
> 
> Yes, NetxPanel overrides the 3 methods that we need the casts for.
> Since they're protected, but NetxPanel is in a different package,
> the compiler gives an error.
> 

Ah okay then, thats what I suspected too.

> > > That said, I wouldn't mind changing the type of appletPanels
> > > to Vector<AVP>. Do you still think I should?
> > >
> > 
> > Actually, perhaps we should go one step up and just change it to
> > NetxPanel? We only store those in the vector anyway.
> 
> We could, but I would rather not because we only use that vector
> for the "applet" field of the AppletPanel class. We don't need
> a NetxPanel for that. Using a NetxPanel would add irrelevant details.
> 

It is not irrelevant to strongly type a container. Strong typing will
catch errors at compile time whereas weak typing will show them at
runtime.

Cheers,
Deepak

> Regards,
> Denis.
> 
> ----- Original Message -----
> > * Denis Lila <dlila at redhat.com> [2011-04-14 09:45]:
> > > > Instead of doing this, it would be easier to change the type of
> > > > appletPanels to Vector<AppletViewerPanel> , then the casting is
> > > > not
> > > > needed as AVP inherits from AP.
> > >
> > > The casting to AppletViewerPanel is not caused by appletPanels. It
> > > is caused by the "panel" member. It used to be declared as an AVP,
> > > then I changed it to a NetxPanel, which made the 3 new casts
> > > necessary (but it eliminated many more casts than it added).
> > >
> > 
> 
> > 
> 
> > 
> > Cheers,
> > Deepak
> > 
> > > Thank you,
> > > Denis.
> > >
> > > ----- Original Message -----
> > > > * Denis Lila <dlila at redhat.com> [2011-04-13 15:10]:
> > > > > > If a patch akin to
> > > > > > http://hg.openjdk.java.net/jdk6/jdk6/langtools/rev/5c2858bccb3f
> > > > > > goes in, I will not be responsible for my actions.
> > > > >
> > > > > I definitely don't want that ;-).
> > > > >
> > > > > Attached is the second patch. It's mostly cleanup, and it
> > > > > also uses "pav" in destroyApplet, instead of calling
> > > > > get(id) again and again.
> > > > >
> > > > > Ok to push?
> > > > >
> > > >
> > > > ...
> > > > ...
> > > >
> > > > <snip>
> > > > >
> > > > >              PluginDebug.debug("getCachedImageRef() getting img
> > > > >              from
> > > > >              URL = ", url);
> > > > > @@ -910,8 +815,8 @@
> > > > >          SocketPermission panelSp =
> > > > >                  new
> > > > >                  SocketPermission(panel.getCodeBase().getHost(),
> > > > >                  "connect");
> > > > >          synchronized(appletPanels) {
> > > > > - for (Enumeration e = appletPanels.elements();
> > > > > e.hasMoreElements();) {
> > > > > - AppletPanel p = (AppletPanel) e.nextElement();
> > > > > + for (Enumeration<AppletPanel> e = appletPanels.elements();
> > > > > e.hasMoreElements();) {
> > > > > + AppletPanel p = e.nextElement();
> > > > >                  String param = p.getParameter("name");
> > > > >                  if (param != null) {
> > > > >                      param = param.toLowerCase();
> > > > > @@ -1591,15 +1496,15 @@
> > > > >           * at the same time.
> > > > >           */
> > > > >          try {
> > > > > - panel.joinAppletThread();
> > > > > - panel.release();
> > > > > + ((AppletViewerPanel)panel).joinAppletThread();
> > > > > + ((AppletViewerPanel)panel).release();
> > > > >          } catch (InterruptedException e) {
> > > > >              return; // abort the reload
> > > > >          }
> > > > >
> > > > >          AccessController.doPrivileged(new
> > > > >          PrivilegedAction<Void>()
> > > > >          {
> > > > >              public Void run() {
> > > > > - panel.createAppletThread();
> > > > > + ((AppletViewerPanel)panel).createAppletThread();
> > > > >                  return null;
> > > > >              }
> > > >
> > > >
> > >