EVALUATION

The suggested fix test isn't available on bugs.sun.com which makes it hard to understand what was changed to fix this.  Basically there's a new "quiet" directive you can put in .hotspot_compiler which makes all following directives silent.

http://analemma.sfbay.sun.com/net/prt-archiver.sfbay/data/archived_workspaces/main/c2_baseline/2006/20061212211559.never.6500501/workspace/webrevs/webrev-2006.12.12/index.html

Fixed 6500501: wildcarding support wanted for CompilerOracle
Fixed 4650676: Hotspot compiler should hide "exclude" messages
      when .hotspot-compiler present

Currently there are two mechanisms for controlling what gets compiled,
-XX:CompileOnly= and the CompilerOracle.  CompileOnly has some strange
implicit wildcarding that seems more an artifact of the implementation
and the design.  The oracle is precise but doesn't support method
signatures as part of the specifier.  This changes adds support for
optional signatures in the directives to the oracle and supports a
limited amount of wildcarding on the names.  It also moves the
CompileOnly logic out of arguments.cpp into the oracle and adds a new
oracle command "compileonly" with the same semantics as
-XX:CompileOnly= but following the parsing rules of the oracle.

To fix 4650676 a new option "quiet" is added which suppresses output
from the oracle.  I also added support for PrintNMethods with the
oracle option command, so
-XX:CompileCommand=option,java/lang/String.append,PrintNMethods will
print out the NMethod for only that compilation.

http://javaweb.sfbay/~never/webrev/6500501

Reviewed by: jrose, kvn

Fix verified (y/n): y

exercised some oracle and CompileOnly command lines to confirm that
-XX:CompileOnly behaves the same as it used to but the new oracle
options behave consistently.
                                     

SUGGESTED FIX

http://analemma.sfbay.sun.com/net/prt-archiver.sfbay/data/archived_workspaces/main/c2_baseline/2006/20061212211559.never.6500501/workspace/webrevs/webrev-2006.12.12/index.html

Fixed 6500501: wildcarding support wanted for CompilerOracle
Fixed 4650676: Hotspot compiler should hide "exclude" messages
      when .hotspot-compiler present

Currently there are two mechanisms for controlling what gets compiled,
-XX:CompileOnly= and the CompilerOracle.  CompileOnly has some strange
implicit wildcarding that seems more an artifact of the implementation
and the design.  The oracle is precise but doesn't support method
signatures as part of the specifier.  This changes adds support for
optional signatures in the directives to the oracle and supports a
limited amount of wildcarding on the names.  It also moves the
CompileOnly logic out of arguments.cpp into the oracle and adds a new
oracle command "compileonly" with the same semantics as
-XX:CompileOnly= but following the parsing rules of the oracle.

To fix 4650676 a new option "quiet" is added which suppresses output
from the oracle.  I also added support for PrintNMethods with the
oracle option command, so
-XX:CompileCommand=option,java/lang/String.append,PrintNMethods will
print out the NMethod for only that compilation.

http://javaweb.sfbay/~never/webrev/6500501

Reviewed by: jrose, kvn

Fix verified (y/n): y

exercised some oracle and CompileOnly command lines to confirm that
-XX:CompileOnly behaves the same as it used to but the new oracle
options behave consistently.


Build:
                                     

EVALUATION

Re-assigning this to hotspot compiler team to determine what or if
we should address this RFE.
###@###.### 2002-03-11

The bug is filed for 1.3.1 but this appears to be an issue with latest JDK. If
work needs to be done for fixing this RFE in older release, then JPSE team will need to look into this as well based on compiler teams feedback.
                                     

PUBLIC COMMENTS

Hotspot compiler should hide "exclude" messages when .hotspot-compiler present