[GHC] #5442: ghc-pkg unregister --user/--global/--package-conf not normative

[GHC]

#5442: ghc-pkg unregister --user/--global/--package-conf not normative
-------------------------------------+-------------------------------------
              Reporter:  guest       |             Owner:  ezyang
                  Type:  bug         |            Status:  new
              Priority:  normal      |         Milestone:  7.10.1
             Component:  ghc-pkg     |           Version:  7.2.1
            Resolution:              |          Keywords:
Differential Revisions:              |  Operating System:  Unknown/Multiple
          Architecture:              |   Type of failure:  Other
  Unknown/Multiple                   |         Test Case:
            Difficulty:  Unknown     |          Blocking:
            Blocked By:              |
       Related Tickets:              |
-------------------------------------+-------------------------------------
Changes (by ezyang):

 * owner:  bgamari => ezyang


Comment:

 OK, I can confirm that there is something fishy going on here. Here is the
 matrix of behaviors that ghc-pkg implements right now:

 * If P is in the user and global, {{{unregister --user}}} removes user,
 {{{unregister --global}}} removes **USER**, {{{unregister}}} removes user
 * If P is in the user but not global, {{{unregister --user}}} removes
 user, {{{unregister --global}}} removes **USER**, {{{unregister}}} removes
 user
 * If P is in the global but not user, {{{unregister --user}}} removes
 **GLOBAL**, {{{unregister --global}}} removes global, {{{unregister}}}
 removes global

 This seems clearly wrong. The bug seems to me that ghc-pkg calculates the
 database to modify fine, but then ignores it in {{{modifyPackage}}},
 always using the full database stack rather than the database it
 calculated to use. My suggested fix is to switch unregister to use the
 flag package stack, since it is implicitly a read, and then a delete. But
 I don't know if there is any existing tooling that would be broken by this
 fix. I have a patch and test-cases.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/5442#comment:8>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler