On Dec 29, 2009, at 5:00 AM, Pavel Klodin wrote:
Hello Ray,
Roughly speaking, a test suite is a directory structure containing
test sources and binaries, necessary libraries and documentation,
etc. And there must be file 'testsuite.jtt' (or file
'testsuite.html' for backwards compatibility) at the top of this
structure. This is how JavaTest identifies test suites.
Thanks for the reply. It is just odd that if I look, for example, in
the bsd-port directory, I can look in directories that are named
"test" and there are 36220 java sources in them. So, the tests are
distributed with the source, but not the "jtt" files? Why would it be
a good idea to not make the infrastructure as widely available as the
soures? Are too many people wanting to run tests? I find that usually
the opposite is true and putting barriers in front of people who want
to run tests does not seem like a very good idea.....
Why are jtt files not a regular part of the source?
BTW, if you are using JCK with OpenJDK then you (or the company
you're working for) must have signed OCTLA. So, please note for
further possible questions that support for the JCK is primarily
handled through a private mailing list, shared by Sun and OCTLA
licensees.
I have gotten myself enough access to be able to check out sources
from the OpenJDK. And I get tests with that. But not the capability of
running them....
Please also visit our (Conformance Group) page at http://openjdk.java.net/groups/conformance
for more information.
I will. Hopefully the rationale will be clearer. Right now, it seems
that separating the JCK and the source is nonsensical. We'll see.
cheers - ray
-------- Original Message --------
Subject: issues with getting started...
Date: Sun, 27 Dec 2009 12:06:56 -0800
From: Ray Kiddy <r...@ganymede.org>
To: quality-discuss@openjdk.java.net
Hello -
I hate posting newbie-ishly, but I cannot avoid it in this case. I
am having problems with some things that seem really, really
basic. I have tried to read the available docs. Some links to
testing articles on java.net have gone stale. Others seem to leave
steps out. For example, I have tried to follow the steps in http://openjdk.java.net/groups/conformance/docs/JCK6bUsersGuide/html/p13.html
.
This may sound stupid, but I am having problem seeing how a test
suite is identified. If I launch javatest, it asks me which suite
I want to run. Is a suite identified by implementing an interface,
or sub- classing some class? If so, I cannot find it. The UI would
suggest there is a suffix and a file type that identifies test
suites. Or else, how would the picker identify one? Using the file
browser to randomly find something that is acceptable as a test
suite has not worked.
Is there a wiki page somewhere that describes the testing effort?
I did a search from a page on the OpenJDK wiki (http://wikis.sun.com/dosearchsite.action?queryString=test
), and I got these results:
# test-test (test-test)
# Test_Test (test_test)
# .test. (.test.)
# testes (testes)
# test (test)
...
Not helpful. Paging through the rest of the results of this search
does not seem to bring one to anything useful. It seems that the
testing effort which is presumably going on within Sun has not yet
been given very much outside visibility. Or I am missing something
obvious, which is completely possible.
I can build OpenJDK 7 on Mac OS X, and I am interested in seeing
how the tests can be run on this platform. I also have the VM
build on an Ubuntu system, but identifying a test suite is not any
more obvious on any other platform. Really, I am sure I am missing
something which may be obvious to many. I often come to things
with, shall we say, a "unique" perspective. But not seeing
something "obvious" can be useful and I often find that
assumptions inherent in a process can be better documented. It
would be especially helpful if the documentation was on a wki.
Any suggestions?
I have been working with java for quite a while. For example, I was
in the WebObjects team at Apple for almost 10 years and was there
while WO was being ported to java. I have some ideas for testing
the VM that might be interesting. But I obviously do not yet
understand something basic about the approach you all are taking.
So, we'll see.
thanx - ray
