Alexei Zakharov wrote:
Let me rephrase my question a little bit. What should we do in this
particular case when some test passes on J9 and HDK (DRLVM with
BUILD_CFG=release) and fails/crashes on DRLVM with BUILD_CFG=debug? I
have two examples of such tests already.
1) Fix DRLVM.
2) conditionally not execute that test when running DRLVM so that it
completes.
But we don't want to comment out tests in the code, just not run them
for certain environments.
geir
Thanks,
2006/11/9, Alexei Zakharov <[EMAIL PROTECTED]>:
Stepan, Tim,
Well, I am not sure here. All tests passes both on J9 and DLRVM Relase
build (HDK is a RELEASE build is far as I understand). But it crashes
on DRLVM DEBUG - this makes our drlvm guys a bit nervous. As far as I
understand the reason is too complex to be fixed promptly. Probably
the solution I have chosen is not the best one. So I am open for
proposals here. Should we exclude the whole test? In my opinion DRLVM
is not just "another_VM" - it is something we should care about. IMHO.
Thanks,
2006/11/8, Tim Ellison <[EMAIL PROTECTED]>:
> Stepan Mishura wrote:
> > Hi Alexei,
> >
> > Sorry, I don't understand your logic. Is the test case valid? If
there was
> > another bug (for example: "[another_VM][unit] half of classlib
beans tests
> > crashes VM"), would you agree to comment out a half of beans tests?
>
> Thanks Stepan, I had also intended to point out this commit. I agree
> that it is masking a problem elsewhere which should be fixed, and I'd
> also point out that commenting out test code is not the preferred
way to
> exclude a test since it will be very difficult to find the excluded
test
> and include it again. We have been using the build scripts to exclude
> test classes.
