RE: [Ikvm-developers] Re: PB on java socket mapping ?

[<ct...@gu...>]

OK, thx for the integration for the next release, no pb for the date (i
have made a patch in jgroups for the moment).

So after the fix of this pb, i have two others:
1) about object which are externalizable: it seems the write/read are not
symetrics and when we read an object (write before on an ouput stream), w=
e
have not the right data awaiting (i look that data are shift by other dat=
a
((not writing by my object)).

2) as mention in my original post, when reading on inputstream i have
exeption: /usr/share/squirrelmail


i've the feeling that this two points are linked (correlation) and after
some investigation, i ask me if is not in the
implementation/interpretation of the method java.io.InputStream.readFully
and/or java.io.DataInputStream.readFully ? can you say me if it depends o=
f
classpath or ikvm ?

i have read the result of mauve test for classpath and found no warning
for the package java.io (jdk1.4).

Thx a lot

Christophe
PS: good travel Jeroen

> ct...@gu... wrote:
>> I think i have found the pb.
>> The error seems to be in implementation of java.net.ServerSocket in
>> classpath:
>> when we try to bind  an address with a port already used, the
>> classpath return an IOException but sun JVM return a sub-classed
>> exception:  BindException
>>
>> i try to fix this pb in classpath mailing list
>
> This is an IKVM "bug" (not really a bug, because the API docs don't
> specify that ServerSocket should throw a BindException).
>
> I will fix it in the next version, but it'll take a while before that
> gets released, since I'm travelling next week.
>
> Regards,
> Jeroen
>