[Nix-dev] Please help generating data about GHC's non-deterministic library ID bug

[Peter Simons]

Hi folks,

we would like to know how frequently the infamous GHC library id bug [1]
actually occurs. To determine that information, we've prepared a little
experiment at:

  https://github.com/peti/ghc-library-id-bug

The repository contains a Nix default.nix file that when run (with some
parameters) builds 25 copies of a given Haskell library and produces as
output a CSV file containing the library id GHC assigned to each of
them.

If you have Nix installed on i686-linux, x86_64-linux, or x86_64-darwin,
then please follow the instructions at

  https://github.com/peti/ghc-library-id-bug/blob/master/README.md

to run those build on your machine, and send the generated CSV files
back to me via private e-mail.

Please send the file back even if all ids are the same, i.e. even if the
bug did not occur! It's important that we get a realistic sample.

If possible, please run builds for both "text" and "aeson". By default,
the build does 25 iterations, but fast machines should easily cope with
100 iterations, too. The more data we get, the more accurate our results
will be, so please donate your CPU cycles generously. :-)

Best regards,
Peter


[1] https://ghc.haskell.org/trac/ghc/ticket/4012