patch applied (ghc): Fix (yet another) odd interaction between selector thunks and compacting GC

[Bulat Ziganshin]

Hello Simon,

Tuesday, November 14, 2006, 3:32:25 PM, you wrote:

>   * Fix (yet another) odd interaction between selector thunks and compacting GC
>
>   But since it's quite difficult to reproduce the error, I can't be 100%
>   certain it's gone.  I certainly can't reproduce it again after the
>   fix, anyway.

i've variant of my program which in ~20% of runs leads to this bug.
so, when i will download ghc version which includes this patch, i will
give you additional sign whether it works or not

btw, this program just have several threads with rather complex
haskell -> c -> haskell calls, plus allocs many data, and at some point it
starts to frequently call performGC. i think that by adding frequent
calls of performGC at some point to any complex program (say GHC) it
should be possible to raise this bug in rather predictable way

Simon, can you please also look into
http://cvs.haskell.org/trac/ghc/ticket/637 ? Inability to handle
^Breaks makes my program look a bit non-professional, it also affects
ghc itself which can't handle gracious exit on ^Break. for example,
^Breaking Cabal builds frequently leaves zombie ghc.exe sub-processes in
memory


-- 
Best regards,
 Bulat                            mailto:Bulat.Ziganshin at gmail.com