Claus Reinke wrote:
i'm looking at buildbot test results after my patches were applied,
and could use some advice on how to address the issues:
1.from the daily report
macgyver PPC OSX head: pass
New unexpected test failures:
ghci024 1 macgyver PPC OSX head
according to buildbot log, ghci falls over on ':set -package ghc'
is this expected or does it need looking into? meanwhile,
should i mark ghci024 as 'expect_fail_if_platform("ppc-osx-head")'?
|-- :set -package ghc
|
|Stderr:
|package flags have changed, ressetting and loading new packages...
|ghc-6.9.20071107: internal error: unconditional relative branch out of
range: jump island out of range
| (GHC version 6.9.20071107 for powerpc_apple_darwin)
| Please report this as a GHC bug:
http://www.haskell.org/ghc/reportabug
Presumably we weren't testing ":set -package ghc" before, so this is not a
new problem. We need to open a ticket, and use expect_broken() in the
testsuite.
2. not in the daily report (because 'lost' after running tests?),
but the buildbot log for 'x86_64-linux-head-unreg' says
http://darcs.haskell.org/buildbot/head/x86-64%20Linux%20head%20unreg/builds/167/step-runtestsuite/1
| - -fasm-mangling
| + -fno-asm-mangling
can i see this configuration default in the testsuite?
is 'config.unregistered' the thing to use?
config.unregisterised (note spelling)
3. is the testsuite shared between HEAD and STABLE?
yes.
it seems that ghci023-025 appear as failing in STABLE
logs, which is to be expected as the patches they test
have been applied to HEAD only.
shouldn't there be an 'expect_fail_if_compiler_lt('ghc','6.9')'
for any such test, if the suite is indeed shared?
We're planning to merge, which is why I didn't update the testsuite. We
could install an expect_broken() until the merge, but it didn't seem worth
it. If the merge is delayed, then it will make sense to do that.
FWIW, I think we could do a lot to improve the reporting of test failures
in the buildbot status emails. I usually tend to look at the actual logs.
Cheers,
Simon
_______________________________________________
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc
