On Fri, Mar 20, 2009 at 4:52 AM, Sian January <sianjanu...@googlemail.com> wrote: > Any updates on these M9 issues today? I think we still need someone > with VM expertise to reply to Charles and someone to investigate the > AWT/Swing crashes. > > Nathan - are you able to look at the Vista-specific issues, or can we > postpone them? I'm not aware that anyone else on the list has a Vista > system so it would be difficult for anyone else to investigate them.
I'm stuck on the Vista issue. There's a bunch of generated code based on some windows APIs, but I can't find the tool and no else seems to recall it. It came from the Intel contribution with AWT/Swing. It's called the 'nativebridge' tool. http://www.google.com/codesearch/p?hl=en#b-U_roQFU0o/modules/awt/src/main/native/win32wrapper/windows/org_apache_harmony_awt_nativebridge_windows_Win32.h&q=harmony%20generated%20lang:c%20package:apache.org%20package:%22http://svn.apache.org/repos/asf/harmony/enhanced/classlib/trunk%22 > > I'm not keen to keep the code base frozen indefinitely for these > issues unless they are actively being worked on. Either way I think > we need to aim to unfreeze by the end of next week and hopefully start > the vote then too. > > On 19/03/2009, Charles Lee <littlee1...@gmail.com> wrote: >> Hi guys, >> I have done some investigation about the SSLSocketImplTest vm crash. >> >> The reason vm crash because we try to write something into a socket, which >> is reset. >> I have try the test on the IBM vme. The "java.net.SocketException: Broken >> pipe" has raised as expected. But on the drlvm no exception raised but vm >> crash. >> >> Any suggestion about this? >> >> On Wed, Mar 18, 2009 at 11:43 PM, Sian January >> <sianjanu...@googlemail.com>wrote: >> >>> 2009/3/18 Tim Ellison <t.p.elli...@gmail.com>: >>> > Thanks for the review... >>> > >>> > Sian January wrote: >>> >> The patch looks fine for this case, but if I was being picky I would >>> >> wonder if we also need to throw an IOException for an infinite stream >>> >> that does contain '\n' characters as well as for one that doesn't? >>> > >>> > The RI runs forever on a stream of '\n's and we run up to an >>> > OutOfMemoryException. It's such a contrived case that I really don't >>> > think we need to hang forever too to match the behavior. >>> > >>> >> Also is '\n' ok to use across all platforms or should it be something >>> >> like System.getProperty(line.separator)? >>> > >>> > The manifest format spec says: >>> > newline : CR LF | LF | CR (not followed by LF) >>> > >>> > so I'll change the containsLine test to be >>> > if (buffer[i] == 0x0A || buffer[i] == 0x0D) { >>> > >>> > Sounds reasonable? >>> >>> Ok - sounds fine >>> >>> +1 for applying >>> >>> > >>> > Thanks again, >>> > Tim >>> > >>> > >>> >> 2009/3/18 Tim Ellison <t.p.elli...@gmail.com>: >>> >>> Sian January wrote: >>> >>>> 2. org.apache.harmony.archive.tests.java.util.jar.ManifestTest fails >>> >>>> on Windows XP [Tim is working on this] >>> >>> I have a proposed patch ready for this problem, see >>> >>> >>> >>> https://issues.apache.org/jira/browse/HARMONY-6121 >>> >>> >>> >>> it's not a one-liner, so I'd appreciate another set of eyeballs on it >>> >>> before nominating it as a commit into M9. >>> >>> >>> >>> Thanks, >>> >>> Tim >>> >>> >>> >>> >>> >> >>> >> >>> >> >>> > >>> >>> >>> >>> -- >>> Unless stated otherwise above: >>> IBM United Kingdom Limited - Registered in England and Wales with number >>> 741598. >>> Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU >>> >> >> >> >> -- >> Yours sincerely, >> Charles Lee >> > > > -- > Unless stated otherwise above: > IBM United Kingdom Limited - Registered in England and Wales with number > 741598. > Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU >