Tim Ellison wrote:
Mikhail Loenko wrote:
Hi George
Wasn't that you who strictly opposed logging to stderr ~3-5 months ago? :)
As I remember, you said that no one will read that stderr.
I said that (George may have too).
Every test either passes or fails. We can agree that skipped tests
count as a pass or a fail, but writing stuff into logs or onto the
console is useless -- nobody, especially the build system, will be
reading output to decide if the 'all tests passed' or 'test failures'
outcome was lying.
Right - if we find a "failure", then we can go look at test or server
logs...
I think the tests we have here are kind of needing 'exotic' configuration,
and they should be separated and not included to the regular pre-commit
run.
The 'exotic' configurations are those that are external to the system
under test and associated test infrastructure.
AIUI George is proposing embedding Jetty into the test framework.
+1
Regards,
Tim
2006/5/19, George Harley <[EMAIL PROTECTED]>:
Hi Stepan,
Yes, there is probably scope for adding some logic to this test method
where - in the event of no network connection we write a message to
stderr and move on. An alternative is to provide for connections to be
opened up against a server on the local machine. A couple of days ago I
spent some time adding the small, lightweight (and Apache licensed)
Jetty server to the LUNI tests. In my local sandbox the LUNI
make/build.xml file the run.tests target was updated to start up Jetty
--> run the LUNI tests --> stop Jetty. It worked fine apart from a
glitch I had getting the cgi-bin to work properly in Jetty. Apparently
this is a fairly common problem when it is used on Windows. I'm starting
to feel that it might be best to exclude the one or two cgi-bin related
tests in the LUNI tests for now in the interests of bringing Jetty on
board. It would certainly help get some more java.net tests up and
running and enable this particular test case to run without the need for
a network connection. What do you think ?
I would, of course, float the Jetty proposal on the dev list first to
make sure we were all happy with the suggestion :-)
Best regards,
George
Stepan Mishura wrote:
Richard, George
I'm not fan of tests that depend on network connection. Is it
necessary for
this test:
+ public void test_getOutputStream_afterConnection() throws Exception {
+ URLConnection uc = new
URL("http://www.apache.org";).openConnection<http://www.apache.org";).openconnection/>
();
+ uc.setDoOutput(true);
+ uc.connect();
+ assertNotNull(uc.getOutputStream());
+ }
Thanks,
Stepan Mishura
------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
