2

I have a stack yaml with resolver lts-9.9 which is associated with ghc 8.0.2 (see https://www.stackage.org/lts-9.9 but when I run stack solver I see:

Using resolver: lts-9.9
Using compiler: ghc-8.0.1

and not ghc-8.0.2. This seems to violate that a stack build is constrained to exactly one set of packages and ghc? What do I overlook?

duplode
  • 33,731
  • 7
  • 79
  • 150
user855443
  • 2,596
  • 3
  • 25
  • 37
  • 1
    I was wondering whether the [`compiler-check` option](https://docs.haskellstack.org/en/stable/yaml_configuration/#compiler-check) might be at play here, but even the `newer-minor` variant shouldn't allow this. – sjakobi Sep 01 '18 at 18:37
  • 1
    thanks - the compiler check helped. I wonder why exact match is not default; a miss on the compiler makesa package compiling on one machine not compiling on another - which is what stack wants to achieve. – user855443 Sep 01 '18 at 19:55
  • So the documentation is wrong then? Could you maybe report this on `stack`'s issue tracker? – sjakobi Sep 02 '18 at 02:33

1 Answers1

0

This can happen if ghc-8.0.1 is a system-based installed version of ghc, and stack is configured to use it instead of its own sandboxed version.

There is a pretty comprehensive page describing all the cases - https://docs.haskellstack.org/en/stable/faq/

ДМИТРИЙ МАЛИКОВ
  • 21,474
  • 11
  • 78
  • 131