OK. One more time.
http://cr.openjdk.java.net/~jgish/Bug7159567-set-logging-MemoryHandler
I compromised with the RuntimeException. I'm instead catching it, but
throwing a new one this way:
65 throw new RuntimeException(
66 "Test Failed: did not load java.util.logging.ConsoleHandler as
expected",
67 rte);
That way, we retain the original, but have the advantage of having a clear indication of
"Test Failed" and the reason. Otherwise, diagnosing
the failure forces the reader to dig into the stack trace.
Thanks,
Jim
On 10/24/2012 08:40 PM, Mandy Chung wrote:
On 10/24/2012 12:31 PM, Jim Gish wrote:
See updated webrev:
http://cr.openjdk.java.net/~jgish/Bug7159567-set-logging-MemoryHandler
Looks good. Thanks for the update.
MemoryHandlerTest.java - thanks for renaming it but you forget to
change L28 @run tag. jtreg should fail if you run this new test.
L64-66 this try-catch block isn't necessary, as I mentioned in my
previous comment, but no big deal if you want to leave it there. The
comment lines and some throw statements are really long and should be
broken into multiple lines (I didn't notice the long lines in previous
versions - sorry if I had missed them). Hopefully it's just one-click
reformat for you using IDE :)
Mandy
Thanks,
Jim
On 10/17/2012 03:46 PM, Mandy Chung wrote:
Hi Jim,
On 10/11/2012 2:37 PM, Jim Gish wrote:
Please review the updated changes at
http://cr.openjdk.java.net/~jgish/Bug7159567-set-logging-MemoryHandler/
The spec change looks good. As Alan points out, </li> is missing.
Although they were not there before, I would think it's a good clean
up while you are in these files if you agree.
Done
The test looks better. Is SimpleTargetHandler.numPublished
intended to be checked? SimpleTargetHandler is set as the target
for java.util.logging.MemoryHandler. I guess you want to create a
logger using the standard MemoryHandler.
Nit: the test is named MemoryHandler and I guess the name conflict
causes the custom handler classes to extend
"java.util.logging.MemoryHandler" with a fully-qualified class
name. As the properties file is named MemoryHandlerTest.props, do
you consider renaming the test to MemoryHandlerTest to avoid the
name conflict? I don't have strong opinion and just want to point
that out.
I don't see this as a problem. However, I've renamed MemoryHandler
to MemoryHandlerTest for improved clarity.
L62-64: better not to rethrow a new RuntimeException as the
exception and stack trace will help diagnostics if it does go
wrong. You can get rid of this try-catch block.
OK -- the reason I did this was to insert a readable message into the
new RuntimeException to indicate the cause of the failure. I think
this is good practice and leads to easier diagnosis, but since you
disagree, I'll take it out.
L120: is it a leftover debug statement? I think you meant to add
test case to exercise this target handler, right?
removed and a few tests added.
....Jim
I've changed as you've requested, added some additional tests, did
some better error handling in the case of a MemoryHandler not
specifying a target (now throws RuntimeException with an
appropriate message instead of attempting to load a null class and
throwing NPE). I also corrected the copyrights, tested with JCK,
all jdk_lang tests and have submitted a JPRT job with core tests.
Great. Thanks for doing it.
Mandy
I've forwarded a CCC request (separately) and will await its
approval and further review of this change.
Thanks,
Jim
On 09/28/2012 05:32 PM, Mandy Chung wrote:
On 9/28/2012 12:13 PM, Jim Gish wrote:
I've re-spun the change with additional usage notes in the spec
to reflect the long-standing actual behavior. Note that it
doesn't change the spec per se, as it was already stated in
LogManager. This change simply replicates that language with an
example in each *Handler class to make it easier to find.
Thanks for looking into it. This statement in LogManager does
specify the properties for handlers:
The properties for loggers and Handlers will have names starting
with the dot-separated name for the handler or logger.
Replicating that statement with an example is one way to do it.
Would it be clearer if the prefix of the properties referenced
in the bullet list is replaced from "java.util.logging" to
some kind of prefix - something like this:
*<b>Configuration:</b>
* By default each<tt>ConsoleHandler</tt> is initialized using
the following
*<tt>LogManager</tt> configuration properties. If properties are
not defined
* (or have invalid values) then the specified default values are
used.
*<ul>
*<li> <handler's classname>.level
* specifies the default level for the<tt>Handler</tt>
* (defaults to<tt>Level.INFO</tt>).
...<snip>
*</ul>
*
* For example, the properties for {@code ConsoleHandler} would be:
* java.util.logging.ConsoleHandler.level=INFO
*
java.util.logging.ConsoleHandler.formatter=java.util.logging.SimpleFormatter
*
* For a custom handler, e.g. com.foo.MyHandler, the properties
would be:
* com.foo.MyHandler.level=INFO
* com.foo.MyHandler.formatter=java.util.logging.SimpleFormatter
This might add some clarity to the spec.
This is a spec bug fix that I would suggest to file a CCC to
track for compatibility. I would also suggest running the JCK
tests to find out if there is any regression due to this fix.
The webrev, as posted at
http://cr.openjdk.java.net/~jgish/Bug7159567-set-logging-MemoryHandler/
See my comment above w.r.t. the spec change.
test/java/util/logging/MemoryHandler.java
L27: "via via" typo
L28: @run tag specifies the test name
So it should be @run main/othervm MemoryHandler
L43: jtreg runs the test in a different working directory
other than the test source. So the test has to read
the system property ("test.src") to determine the location
of the properties file. Typically, we will do this:
String src = System.getProperty("test.src", ".);
File fname = new File(src, LM_PROP_FNAME);
You don't need L44. You can reference LoggingDeadlock3.java test.
L51: this catch block to throw a RuntimeException doesn't seem
necessary. If NPE is thrown, the test will fail anyway.
One suggestion to the test is to test both cases (one with
the specified target handler and one without). You can
define a custom target handler so that the test can verify
if the expected one is used. A simple handler to count
the number of log messages will do the work.
test/java/util/logging/MemoryHandlerTest.props
I suggest to take out the comments and just keep the
properties the test needs to make it easier to tell
what's configured.
Perhaps you should also specify
java.util.logging.MemoryHandler.target to make sure
that the custom handler with no target handler specified
will not use j.u.l.MemoryHandler.target as the default.
Mandy
--
Jim Gish | Consulting Member of Technical Staff | +1.781.442.0304
Oracle Java Platform Group | Core Libraries Team
35 Network Drive
Burlington, MA 01803
jim.g...@oracle.com
