Hello Claus,

Claus Reinke <[EMAIL PROTECTED]> wrote:

> i'm surprised to see ghci024 failing unexpectedly on
> "tnaur PPC OSX head". the reports never seem to
> be in visible range when i look at
> 
> http://darcs.haskell.org/buildbot/head/

This is probably because I these days mostly manage to run the tnaur PPC OSX 
builds only in the weekend. If you try Saturday or Sunday, you should be able 
to see the builds.

> 
> but one can reconstruct the last build url (could there
> please be a direct link from the buildbot page headers
> to the last builds?)
> 
> did i misspell the platform in ghci/scripts/all.T
> (the spelling matches the error message, but
> various different spellings seem to be in use), or
> does expect_broken_if_platform not work if
> composed with prepare024?
> 
> all.T:
> 
> execfile(in_testdir('ghci024.py')) # load function prepare024
> test('ghci024', compose(prepare024,
> expect_broken_if_platform(1845,"powerpc_apple_darwin")),
> ghci_script, ['ghci024.script'])
> 
> 
> testsuite log (build 167):
> 
> ghc-6.9.20080118: internal error: unconditional relative branch out
> of range: jump island out of
> range
> (GHC version 6.9.20080118 for powerpc_apple_darwin)
> Please report this as a GHC bug:
> http://www.haskell.org/ghc/reportabug
> 
> *** unexpected failure for ghci024(ghci)
> 

In a recent stable build (which has the same problem), I tried

> test('ghci024', 
> compose(prepare024,expect_broken_if_platform(1845,"powerpc-apple-darwin")), 
> ghci_script, ['ghci024.script'])

(replacing the underscores with dashes) and turned the failure into an expected 
one.

> 
> it would be nice to use an enum type or, if python
> doesn't have that, to check platform parameters
> to expect_broken_if_platform for validity?
> 
> claus

Best regards
Thorkil

_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc

Reply via email to