|Syzygy-||Igloo: I probably mean you.|
|SamB||what is it miscompiling itself to do?|
|Syzygy-||Igloo: There are two builds defined for mnemosyne, but only on ever runs. Is this deliberate?|
|Igloo||Syzygy-: We haven't got around to making them alternate yet|
|fasta||Ok, this is not good. GHC and ghci produce different out comes....|
|sorear||what does -xc say?|
|fasta||sorear: I was only working in ghci "mode" first, I will compile it with the right options now.|
Perhaps you haven't installed the profiling libraries for package EdisonAPI-1.2.1? <- ARG
Build profiling libraries by default, please...
|sorear||oh, that could be a problem ...|
|action||fasta builds edison with profiling libs...|
|fasta||builds edison with profiling libs...|
|sorear||(./Setup.lhs configure -p btw)|
|fasta||sorear: Ok, I ran as you specified. It doesn't give any extra output.|
sorear: in other words: GHC runs the code correctly, ghci doesn't.
|sjanssen||fasta: how does ghc react with -threaded?|
|sorear||(compile time option)|
|sjanssen||ghci uses the threaded RTS iirc|
ghc-6.6: combination not supported: Threaded/Profiling
|sorear||unfortunately -xc only works with profiling (it uses the cost-centre stacks), so you won't get a backtrace|
|sjanssen||so ghci is reporting deadlock, but ghc isn't (with and without -threaded)?|
sjanssen: I am not catching any exceptions.
sjanssen: maybe that matters
sjanssen: The exception only gets thrown when I do C-c C-c at the moment the computation uses zero CPU
|sjanssen||can you paste the code?|
|fasta||sjanssen: no :(|
sjanssen: I could try to produce a simpler instance.
Ok, that's cool. I let ghci segfault
Probably some .o files are reused by ghci incorrectly and then it dies.
|sjanssen||that sounds likely|