#1372: Recompilation checker should consider package versions (and other
factors)
-------------------------+--------------------------------------------------
Reporter: bringert | Owner: simonmar
Type: bug | Status: new
Priority: normal | Milestone: 6.8 branch
Component: Compiler | Version: 6.6
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Os: Unknown | Testcase:
Architecture: Unknown |
-------------------------+--------------------------------------------------
Comment (by simonmar):
Replying to [comment:10 dons]:
I'd like to understand what happened a bit better, because I think there
are actually several different issues. Correct me if I'm wrong, but the
error in the thread you referred to was caused by:
* X11-extras was updated, the API was changed, the version was not bumped
* xmonad was updated to match
* it failed to compile against the old X11-extras
that's expected, right?
And there seems to be some problem with Cabal not noticing when a .hsc
file has been modified, that sounds like a Cabal bug.
The problem you described seems slightly different:
* X11-extras is modified, the version is bumped
* xmonad is updated to match (with a dependency on the new version)
* `darcs pull xmonad`
* without cleaning first, `setup build`
Now, Cabal doesn't notice that `xmonad.cabal` has been updated, and hence
doesn't notice that it needs to depend on the new version of X11-extras.
If this is the case, it's a Cabal bug - Cabal should refuse to build if
the `.cabal` file has been modified since the last configure.
I've been thinking about how to solve the original recompilation problem.
I even hacked up the fix proposed earlier, but then decided it was wrong.
We can't rely on packages being registered to notice when things change -
the compiler really needs to look at the `.hi` files. Ideally, we need to
create a fingerprint of the interface (hence SimonPJs message to the
Haskell mailing list about fingerprints).
None of this will help unless Cabal invokes GHC to do recompilation
checking: so if Cabal starts doing its own dependency analysis, we'll be
in trouble again.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/1372#comment:13>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
