Package for import is guessed incorrectly

[edsko]

It seems that if an import is for a module which is available both in the home package and in a package installed in the system, we might construct the wrong IdInfo (the ghc-error test for issue #88 fails for this reason when the gitlib-cross package is installed, because it exports module "Foo").

[Mikolaj]

I've verified the problem is unrelated to #88 and very minor. I've pushed tests that confirm that. I'm suspending the work on this issue until further orders.

18:45 < mikolaj> from http://www.haskell.org/ghc/docs/7.4.2/html/users_guide/packages.html#package-overlaps
18:45 < mikolaj> It is possible that by using packages you might end up with a program that contains two modules with the same name: perhaps you used a package P that has a hidden module M, and there is also a module M in your program.
Or perhaps the dependencies of packages that you used contain some overlapping modules. Perhaps the program even contains multiple versions of a certain package, due to dependencies from other packages.
18:45 < mikolaj> None of these scenarios gives rise to an error on its own[8], but they may have some interesting consequences. For instance, if you have a type M.T from version 1 of package P, then this is not the same as the type M.T
from version 2 of package P, and GHC will report an error if you try to use one where the other is expected.
18:46 < mikolaj> so it seems it's unspecified which module will be used --- it just happens that our idInfo code picks a different package than GHC API in this case

[edsko]

[18:46] but I am surprised that the test fails, since defOpts includes -hide-all-packages, so Foo from gitlib-cross should not be visible
[18:47] mikolaj: ^^
[18:47] mikolaj: maybe it's enough to fx our code to look at visible packages only?

[Mikolaj]

12:50 < mikolaj> so I added -hide-all-packages to the test
12:50 < edsko> but isn't that what we provide by default?
12:50 < mikolaj> not on master and experimental
12:50 < edsko> ah
12:50 < edsko> ok
12:50 < edsko> so with -hide-all-packages it works?
12:50 < mikolaj> yep
12:50 < edsko> great
12:50 < mikolaj> see my commit

[edsko]

This is due to a TODO in ghc itself; see #95 (comment) for details. (And this is still a TODO in ghc 7.8, incidentally.)