Jeroen Frijters wrote:
Dalibor Topic wrote:
Blake Meike wrote:
Dalibor Topic wrote:
I assume you've run into the same gcc3 vs. gcc4 issue that
Kelvin ran
into with 1.1.7. It's been fixed in CVS, indeed.
I spoke too soon. Everything seems to be working just fine, but two
of the Unit Tests failed:
FAIL: TestSerialVersions.java
FAIL: TestSerialPersistent.java
I'm including the failure logs, in case you are interested.
Right, that's a failure that was introduced when I merged in
the latest
classpath code.
The failures were apparently introduced with Jeroen's patch from
2006-08-11 Jeroen Frijters <jeroen@xxxxxxxxxxxx
<mailto:jeroen@xxxxxxxxxxxx>>
* java/io/ObjectInputStream.java (readClassDescriptor):
Use class's class loader to resolve field types.
* java/io/ObjectStreamField.java
(ObjectStreamField(String,String,ClassLoader)): Removed.
(ObjectStreamField(String,String)): Don't try to
resolve typename.
(resolveType): New method.
but I've not investigated further yet, after I saw that the expected
output for the test cases did not match what
jdk 1.6 outputs for them, anyway, and the output I got with jamvm &
classpath did not match jdk1.6 output
either.
I can't see how that patch could have caused this (but given the complexity of the code I certainly don't rule anything out).
Yeah, it was just the patch after I noticed it, which may mean that the
test is partly bogus, for example.
I've attached a hack that makes the two mentioned tests pass again (and I'm reasonably confident that won't cause any regressions), but it is not the right way to fix this problem. However, the truth of the matter is that IMO our serialization code is broken beyond repair, so I don't think it's worth it to try to fix it the right way. We should just merge in the Sun serialization code once that is available ;-)
Indeed. :)
Thank you very much for the hack, I'll apply it in Kaffe. I've been
hesitant to dive into serialization code myself for the same reason you
mention ...
cheers,
dalibor topic
