You are viewing a plain text version of this content. The canonical link for it is here.
Posted to derby-dev@db.apache.org by Mike Matrigali <mi...@sbcglobal.net> on 2006/06/16 00:12:20 UTC
Did the MessageBundleTest pass before it was made part of derbyall?
From the tinderbox report it seems to have failed from the time the
following change was committed:
Changes from 414231/414413 to 414416:
------------------------------------------------------------------------
r414416 | davidvc | 2006-06-15 02:00:44 +0200 (Thu, 15 Jun 2006) | 4 lines
Removed SQL States and MessageId message ids that are unused and which
don't have matching messages, and then made MessageBundleTest a
part of derbyall.
------------------------------------------------------------------------
U java/shared/org/apache/derby/shared/common/reference/SQLState.java
U java/shared/org/apache/derby/shared/common/reference/MessageId.java
U
java/testing/org/apache/derbyTesting/functionTests/tests/i18n/MessageBundleTest.java
A
java/testing/org/apache/derbyTesting/functionTests/master/MessageBundleTest.out
U
java/testing/org/apache/derbyTesting/functionTests/suites/i18nTest.runall
U tools/testing/i18nTestGenerator/genClient2.sed
Re: Did the MessageBundleTest pass before it was made part of derbyall?
Posted by David Van Couvering <Da...@Sun.COM>.
Yes, I understand, thanks.
David
Andrew McIntyre wrote:
> On 6/16/06, David Van Couvering <Da...@sun.com> wrote:
>> Hm, thanks Andrew. That may very well be, but it was also the case that
>> the master file did not get correctly checked in. I was running in the
>> classes directory. I'll run with the jar files, and see what happens,
>> and correct my logic as needed.
>>
>> Meanwhile, I'll remove MessageBundleTest from derbyall.
>
> Note that the test will pass when running from the classes because
> SQLState will be in your classpath as part of the full class tree. It
> will continue to fail in jar testing because of the problem I
> described above.
>
> andrew
Re: Did the MessageBundleTest pass before it was made part of derbyall?
Posted by Andrew McIntyre <mc...@gmail.com>.
On 6/16/06, David Van Couvering <Da...@sun.com> wrote:
> Hm, thanks Andrew. That may very well be, but it was also the case that
> the master file did not get correctly checked in. I was running in the
> classes directory. I'll run with the jar files, and see what happens,
> and correct my logic as needed.
>
> Meanwhile, I'll remove MessageBundleTest from derbyall.
Note that the test will pass when running from the classes because
SQLState will be in your classpath as part of the full class tree. It
will continue to fail in jar testing because of the problem I
described above.
andrew
Re: Did the MessageBundleTest pass before it was made part of derbyall?
Posted by David Van Couvering <Da...@Sun.COM>.
Hm, thanks Andrew. That may very well be, but it was also the case that
the master file did not get correctly checked in. I was running in the
classes directory. I'll run with the jar files, and see what happens,
and correct my logic as needed.
Meanwhile, I'll remove MessageBundleTest from derbyall.
David
Andrew McIntyre wrote:
> On 6/15/06, David Van Couvering <Da...@sun.com> wrote:
>> I ran derbyall on my machine prior to checkin and it worked. But this
>> checkin was the first time derbyall included MessageBundleTest. It did
>> not pass before then. I must have not properly checked in the output
>> file, is all I can think of. I'll look into this, apologies.
>
> The problem is that MessageBundleTest has a direct reference to the
> class file for the SQLState class
> (org.apache.derby.shared.common.reference.SQLState.class), which,
> during compilation, compiles in a reference to the actual class and
> not one of the public final strings defined by the SQLState class.
>
> During compilation, the SQLState class is normally factored out by the
> compiler, since the Strings it contains can be placed directly into
> the constant pool of classes that reference it.
>
> However, since MessageBundleTest contains a direct reference to
> SQLState.class, a reference to the fully qualified class file
> SQLState.class remains in MessageBundleTest.class. This means that the
> class file for SQLState needs to be present at runtime (e.g., in
> derbyTesting.jar), or the direct reference to SQLState.class in
> MessageBundleTest:54 needs to be removed, for the test to pass.
>
> andrew
Re: Did the MessageBundleTest pass before it was made part of derbyall?
Posted by Andrew McIntyre <mc...@gmail.com>.
On 6/15/06, David Van Couvering <Da...@sun.com> wrote:
> I ran derbyall on my machine prior to checkin and it worked. But this
> checkin was the first time derbyall included MessageBundleTest. It did
> not pass before then. I must have not properly checked in the output
> file, is all I can think of. I'll look into this, apologies.
The problem is that MessageBundleTest has a direct reference to the
class file for the SQLState class
(org.apache.derby.shared.common.reference.SQLState.class), which,
during compilation, compiles in a reference to the actual class and
not one of the public final strings defined by the SQLState class.
During compilation, the SQLState class is normally factored out by the
compiler, since the Strings it contains can be placed directly into
the constant pool of classes that reference it.
However, since MessageBundleTest contains a direct reference to
SQLState.class, a reference to the fully qualified class file
SQLState.class remains in MessageBundleTest.class. This means that the
class file for SQLState needs to be present at runtime (e.g., in
derbyTesting.jar), or the direct reference to SQLState.class in
MessageBundleTest:54 needs to be removed, for the test to pass.
andrew
Re: Did the MessageBundleTest pass before it was made part of derbyall?
Posted by David Van Couvering <Da...@Sun.COM>.
I ran derbyall on my machine prior to checkin and it worked. But this
checkin was the first time derbyall included MessageBundleTest. It did
not pass before then. I must have not properly checked in the output
file, is all I can think of. I'll look into this, apologies.
David
Mike Matrigali wrote:
> From the tinderbox report it seems to have failed from the time the
> following change was committed:
>
>
> Changes from 414231/414413 to 414416:
> ------------------------------------------------------------------------
> r414416 | davidvc | 2006-06-15 02:00:44 +0200 (Thu, 15 Jun 2006) | 4 lines
>
> Removed SQL States and MessageId message ids that are unused and which
> don't have matching messages, and then made MessageBundleTest a
> part of derbyall.
>
> ------------------------------------------------------------------------
>
> U java/shared/org/apache/derby/shared/common/reference/SQLState.java
> U java/shared/org/apache/derby/shared/common/reference/MessageId.java
> U
> java/testing/org/apache/derbyTesting/functionTests/tests/i18n/MessageBundleTest.java
>
> A
> java/testing/org/apache/derbyTesting/functionTests/master/MessageBundleTest.out
>
> U java/testing/org/apache/derbyTesting/functionTests/suites/i18nTest.runall
> U tools/testing/i18nTestGenerator/genClient2.sed
>