Mylar and NLS in rawhide

[Andrew Overholt]

Hi,

eclipse-mylar and eclipse-sdk-nls have finally hit rawhide!  They should
probably show up on mirrors sometime tomorrow.

Thanks to Kyu Lee who did all the work on the translations.  He just
showed me the SDK running with the Korean translations and it's pretty
awesome.

As for Mylar, I'm aware of class library issues with gcj and I'm
tracking them down now.  Filing bugs would help with this as I'm sure I
won't exercise all code paths.  Thanks to Anthony and others who helped
make this happen.

Andrew

[Andrew Overholt]

* Andrew Overholt  [2007-03-20 17:18]:
> As for Mylar, I'm aware of class library issues with gcj and I'm
> tracking them down now.

Here's one example of not being able to add a bugzilla task repository
(https://bugs.eclipse.org in this case):

java.lang.reflect.InvocationTargetException
   at org.eclipse.mylar.internal.bugzilla.ui.tasklist.BugzillaRepositorySettingsPage$3.run(BugzillaRepositorySettingsPage.java:210)
   at org.eclipse.jface.operation.ModalContext$ModalContextThread.run(org.eclipse.jface_3.2.2.M20061214-1200.jar.so)
Caused by: java.lang.UnsupportedOperationException
   at javax.net.SocketFactory.createSocket(libgcj.so.8rh)
   at org.eclipse.mylar.internal.tasks.core.SslProtocolSocketFactory.createSocket(SslProtocolSocketFactory.java:101)
   at org.apache.commons.httpclient.HttpConnection.open(HttpConnection.java:706)
   at org.apache.commons.httpclient.MultiThreadedHttpConnectionManager$HttpConnectionAdapter.open(MultiThreadedHttpConnectionManager.java:1321)
   at org.apache.commons.httpclient.HttpMethodDirector.executeWithRetry(HttpMethodDirector.java:386)
   at org.apache.commons.httpclient.HttpMethodDirector.executeMethod(HttpMethodDirector.java:170)
   at org.apache.commons.httpclient.HttpClient.executeMethod(HttpClient.java:396)
   at org.apache.commons.httpclient.HttpClient.executeMethod(HttpClient.java:324)
   at org.eclipse.mylar.internal.bugzilla.core.BugzillaClient.connectInternal(BugzillaClient.java:261)
   at org.eclipse.mylar.internal.bugzilla.core.BugzillaClient.getConnect(BugzillaClient.java:221)
   at org.eclipse.mylar.internal.bugzilla.core.BugzillaClient.logout(BugzillaClient.java:292)
   at org.eclipse.mylar.internal.bugzilla.core.BugzillaClient.validate(BugzillaClient.java:206)
   at org.eclipse.mylar.internal.bugzilla.ui.tasklist.BugzillaRepositorySettingsPage$3.run(BugzillaRepositorySettingsPage.java:201)
   ...1 more

Andrew

[Andrew Haley]

Andrew Overholt writes:
> Andrew Overholt  [2007-03-20 17:18]:
> > As for Mylar, I'm aware of class library issues with gcj and I'm
> > tracking them down now.
>
> Here's one example of not being able to add a bugzilla task repository
> (https://bugs.eclipse.org in this case):

Without knowing the real class at
javax.net.SocketFactory.createSocket(), no-one has a chance of
figuring this out.

Andrew.

[Andrew Overholt]

* Andrew Haley  [2007-03-21 07:05]:
> Andrew Overholt writes:
> > Andrew Overholt  [2007-03-20 17:18]:
> > > As for Mylar, I'm aware of class library issues with gcj and I'm
> > > tracking them down now.
> >
> > Here's one example of not being able to add a bugzilla task repository
> > (https://bugs.eclipse.org in this case):
>
> Without knowing the real class at
> javax.net.SocketFactory.createSocket(), no-one has a chance of
> figuring this out.

Sorry, I didn't mean to imply that someone should be able to.  I plan to
create a small test case today and file it.

Andrew

[Mark Wielaard]

On Wed, 2007-03-21 at 08:55 -0400, Andrew Overholt wrote:
> Sorry, I didn't mean to imply that someone should be able to.  I plan to
> create a small test case today and file it.

BTW. For those playing along at home, Andrew created
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31302 with a small test
case, that has a fixed for it checked into classpath CVS by Casey now.
Needs backporting to gcj and retesting now.

Cheers,

Mark

[Mark Wielaard]

On Fri, 2007-03-23 at 10:44 +0100, Mark Wielaard wrote:
> On Wed, 2007-03-21 at 08:55 -0400, Andrew Overholt wrote:
> > Sorry, I didn't mean to imply that someone should be able to.  I plan to
> > create a small test case today and file it.
>
> BTW. For those playing along at home, Andrew created
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31302 with a small test
> case, that has a fixed for it checked into classpath CVS by Casey now.
> Needs backporting to gcj and retesting now.

Wait, that is actually wrong. The fix that got checked into classpath
CVS is for "FYI: fix keyUsage bit checking", the patch mentioned in the
bug report isn't in CVS yet (waiting for more testing).

Cheers,

Mark

[Tom Tromey]

> > > > > "Mark" == Mark Wielaard  writes:

Mark> Wait, that is actually wrong. The fix that got checked into classpath
Mark> CVS is for "FYI: fix keyUsage bit checking", the patch mentioned in the
Mark> bug report isn't in CVS yet (waiting for more testing).

I was wondering if we need or want that keyUsage patch as well.
Do you know?

Tom

[Tom Tromey]

Following up my own post.

Mark> Wait, that is actually wrong. The fix that got checked into classpath
Mark> CVS is for "FYI: fix keyUsage bit checking", the patch mentioned in the
Mark> bug report isn't in CVS yet (waiting for more testing).

Tom> I was wondering if we need or want that keyUsage patch as well.
Tom> Do you know?

Casey seemed to think this was reasonably important, so I'm going to
pull it in.

BTW, I think I have caught up on the patch backlog.  At least, all the
important patches that have been pointed out to me and that I didn't
somehow lose have gone in.

If you know of a gcj patch needing attention, let me know and I will
get it in.

Meanwhile I suppose I will go back to trying to fix pending bugs.

Tom

[Andrew Overholt]

* Tom Tromey  [2007-03-28 12:41]:
> BTW, I think I have caught up on the patch backlog.  At least, all the
> important patches that have been pointed out to me and that I didn't
> somehow lose have gone in.

Jakub:  can we request a new RPM spin at your earliest convenience?

Thanks,

Andrew