[icedtea-web] RFC: Patch to use wait() when waiting for conditions to become true

[Deepak Bhole]

* Omair Majid <omajid at redhat.com> [2011-04-14 15:00]:
> On 04/14/2011 02:29 PM, Deepak Bhole wrote:
> >* Deepak Bhole<dbhole at redhat.com>  [2011-04-14 11:05]:
> >>>  * Omair Majid<omajid at redhat.com>  [2011-04-14 10:56]:
> >>>>  >  On 04/14/2011 01:56 AM, Deepak Bhole wrote:
> ><snip>
> >>>
> >>>  Ah, neat. I didn't know that. I will make the change.
> >>>
> >New patch attached. It uses nanoTime().
> >
> 
> Just one concern, noted below.
> 
> >@@ -142,14 +140,19 @@
> >          initEventQueue(panel);
> >
> >          // Wait for panel to come alive
> >-        int maxWait = PluginAppletViewer.APPLET_TIMEOUT; // wait for panel to come alive
> >-        int wait = 0;
> >-        while (!panel.isAlive()&&  wait<  maxWait) {
> >-            try {
> >-                Thread.sleep(50);
> >-                wait += 50;
> >-            } catch (InterruptedException ie) {
> >-                // just wait
> >+        // Wait implemented the long way so that
> >+        // PluginAppletViewer.waitTillTimeout() needn't be exposed.
> >+        long maxTimeToSleep = PluginAppletViewer.APPLET_TIMEOUT;
> >+
> >+        synchronized(panel) {
> >+            while (!panel.isAlive()&&  maxTimeToSleep>  0) {
> >+                long sleepStart = System.nanoTime();
> >+
> >+                try {
> >+                    panel.wait(maxTimeToSleep);
> >+                } catch (InterruptedException e) {} // Just loop back
> >+
> >+                maxTimeToSleep -= System.nanoTime() - sleepStart;
> 
> maxTimeToSleep is in ns, but wait expects ms. waitTillTimeout()
> converts the timeout in ns to ms before passing it to wait(). I
> think we should do something similar here.
>

Ah, good catch. I will add a conversion there before committing.
 
> Everything else looks fine to me.
> 

Thanks!

Deepak

> Cheers,
> Omair