[kaffe] [Bug Report] Kaffe 1.1.4 on GNU/Linux PowerPC: 2 of 144 tests failed

[Dalibor Topic]

Hi Luca,

Luca Saiu wrote:
>   Hi. I've compiled Kaffe 1.1.4 on my GNU/Linux PowerPC machine.
> I used the just realeased GCC 3.3.3. I ran configure without parameters.
> 
>   Two tests failed, but for what I undestand they seem more like bugs in 
> the testsuite apparatus than actual problems in Kaffe:
> 
>   The first failed test is TestSerialVersions. The behaviour is exactly 
> the same I obtain compiling to native code with GCJ or using the GCJ JVM 
> gij (either from bytecode generated by gcj or by the Kaffe compiler, 
> nothing seems to change).
>   The output is (only):
> -------------------------------------------
> Generating canonical stream
>   Serializing TestSerialVersions$Test0001
>     OK.
> -------------------------------------------
>   The comment at the end of the source file says it should be longer.

This was a bug in our and GNU Classpath Serialization code. It was fixed 
after 1.1.4 was released by Guilhem.

>   The other failed test is GCTest. However I've run it several times, 
> and the output, after it has been sorted, seems to be always identical 
> to the one in the source file comment.

>   I'm not sure there is some real problem, however the output from "make 
> check" told me to report this as a problem, so I did :-).

That's weird. It may be some problem with the TestScript machinery. 
Unfortunately, it works fine for me here on i386-linux, so I can't 
reproduce it. :(

Maybe there is some whitespace difference in the output?

cheers,
dalibor topic