So I have two replies from Simon and Ian suggesting a minimal behind the scenes patch for QSem (possible) or splitting them out of base. I have considered both of these. Over the releases of GHC there have been several bits separated out of base, each time requiring another round of cabal file updating. If people want to carve more out of base this is an independent decision to updating QSem/QSemN/SampleVar. Thanks for looking through the code, Ian, and finding where it was used. I'll go look at that to see how safe the usage is. With this encouragement, I'll prepare a patch to GHC to replace the guts of the modules and update the documentation. I'll patch against ghc-7.0.3, and I don't expect conflicts on the these on files. Cheers, Chris