Re: updating an ImageIcon's backing BufferedImage content from multiple threads

From:

Knute Johnson <nospam@rabbitbrush.frazmtn.com>

Newsgroups:

comp.lang.java.programmer

Date:

Sun, 05 Apr 2009 21:39:57 -0700

Message-ID:

<49d9878e$0$19532$b9f67a60@news.newsdemon.com>

Lew wrote:

Knute Johnson wrote:

For example, draw on a BufferedImage in a worker thread, and then
display that BufferedImage by calling repaint(). That would change
pixels in one thread and display in the EDT.

I worry that the change to the BI in one thread is not necessarily
visible to the other unless you synchronize the accesses to the BI.

On the other hand, I've never seen a case where a change was not
visible from one thread to another. I know that there is no
guarantee that it will be so

That fuels my concern.

but I have never seen a case of it with Sun's JVM. It is not
required to be invisible, it just may be. I would really like to be
able to create a program that could demonstrate a lack of visibility
across threads.

Never having seen the bug is not proof that it is not lurking. Sun went
for many years thinking that one could construct GUIs off the EDT and
then use them on the EDT, but now admit that was wrong. Those bugs were
not observed for a long, long time.

The thing is, it may be thread safe because of something in Sun's code
and that is why we can't make a visibility problem occur. All it would
take in this case is for repaint() to write to a volatile variable and
the EDT to read from it and all other variables would be visible.

--

Knute Johnson
email s/nospam/knute2009/

--
Posted via NewsDemon.com - Premium Uncensored Newsgroup Service
------->>>>>>http://www.NewsDemon.com<<<<<<------
Unlimited Access, Anonymous Accounts, Uncensored Broadband Access