Hello Linux JDK Porters,
1.1.6v5 works great for us but here is a small bug report.
(I use this channel because Jitterbug gave me
Forbidden
You don't have permission to access /cgi-bin/jdk on this server.
)
1. Startup is slowed down by factor of 2.5 (!) by checkVersions.
The following works for me:
If you get a working "java" but starts up much more slowly
than previous versions you may take a look at
/usr/local/java-1.1.6v5/bin/checkVersions. This scripts checks
for incompatible shared libraries on your systems and decides
wether to use the JDK ones.
Obviously (:-) it is not necessary to do that on every startup
if you know the answer.
So to check what checkVersions thinks about your needs,
enter
sh -x java
in your shell.
Look for the lines
+ '[' -x /usr/local.local/java-1.1.6v5/bin/checkVersions ']'
+ /usr/local.local/java-1.1.6v5/bin/checkVersions java
if the next line reads
JDK_LIBS="<something>/linuxlibs:"
you may try to set (csh syntax)
setenv JDK_NO_VERS_CHECK 1
otherwise
setenv JDK_NO_JDK_LIBS 1
to avoid the version check on startup.
The necessary diff on checkVersions is appended.
A tip when using tya (a nice easy to install and use JIT for Linux,
give it a try for speeding up your compile times):
Java itself worked for me with both settings above,
but when using "setenv JDK_NO_VERS_CHECK 1"
java gives
SIGSEGV 11* segmentation violation
stackbase=0xbfffe570, stackpointer=0xbfffd904
heyl@alexx 62: rcsdiff -c /usr/local/java-1.1.6v5/bin/checkVersions
===================================================================
RCS file: /usr/local/java-1.1.6v5/bin/checkVersions,v
retrieving revision 1.1
diff -c -r1.1 /usr/local/java-1.1.6v5/bin/checkVersions
*** /usr/local/java-1.1.6v5/bin/checkVersions 1998/11/05 14:33:48 1.1
--- /usr/local/java-1.1.6v5/bin/checkVersions 1998/11/05 14:34:34
***************
*** 226,231 ****
--- 226,235 ----
then
exit 0
fi
+ if [ "$JDK_NO_JDK_LIBS" ]
+ then
+ exit 1
+ fi
computeDirVars
2. The rmi-Compiler (rmic) has a bug which makes it a pain to use
on linux nfs mounted file systems.
It writes it output files "byte-by-byte" so each byte is
travvelling in its own RPC-Packet to the NFS-Server.
You can imagine that this is sloooooooow. :-)
I think adding a BufferedOutputStream (or Writer) should
cure that problem.
Ciao,
Carsten Heyl
------------------------------------------------------------------
Carsten Heyl [EMAIL PROTECTED]
NADS - Advertising on Nets
NADS GmbH http://www.nads.de/
Otto-Hahn-Str. 18 http://www.pixelboxx.de/
D-44227 Dortmund Tel.: +49 231 975 123-0
