Hi,
This is a bug which has been in GHCi from the beginning.
Bug
===
GHCi interprets a module while the compiled version is
present and up-to-date.
Details
===
When I (for example) have the following module structure:
module A where
...
module B where
import A
...
The following thing happens in GHCi:
First, I start GHCi (without any arguments). Then I compile
both modules (I am using the command-line here, but that has
nothing to do with the bug.)
Prelude :!ghc --make B
ghc-5.02.2: chasing modules from: B
Compiling A( A.hs, A.o )
Compiling B( B.hs, ./B.o )
This looks fine. Now, I load B into GHCi:
Prelude :l B
Skipping A( A.hs, A.o )
Skipping B( B.hs, ./B.o )
Ok, modules loaded: B, A.
Both modules are loaded, and it is the compiled version for
both.
Now, I am making changes to A:
B :!touch A.hs
And I recompile the whole thing:
B :!ghc --make B
ghc-5.02.2: chasing modules from: B
Compiling A( A.hs, A.o )
Compiling B( B.hs, ./B.o )
Still everything is fine. However, when I now reload using
:r, GHCi recognizes that A.hs has changed, but it does *not*
recognize that the new A.o file is already there!
B :r
Compiling A( A.hs, interpreted )
Compiling B( B.hs, interpreted )
Ok, modules loaded: B, A.
So, I get the interpreted versions of both! (This is wrong.)
The only way to get the compiled versions, is to load B
freshly:
B :l B
Skipping A( A.hs, A.o )
Skipping B( B.hs, ./B.o )
Ok, modules loaded: B, A.
This leads to the state that I want GHCi to be in.
/Koen.
___
Glasgow-haskell-bugs mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
