Claus Reinke wrote:
* fix intermittent failure with ghci025
M ./tests/ghc-regress/ghci/scripts/ghci025.script +2
-- ToDo: this will go wrong in parallel
:! cp ../prog001/D1.hs ../prog001/D.hs
wouldn't it be easier to enforce the invariant that prog001/* is always
compilable? ie, instead of having to create a D.hs
when needed, always have a D.hs, and just overwrite it
with D[12].hs as needed. that should work in parallel as
well because ghci025 only looks at the types, which don't change.
That wouldn't work, you could still catch it halfway through copying, for
example.
alternatively, it might be useful if scripts could create or copy the
modules they need on the fly (some already do, i think, but there should
be script-specific temporary
directories to put source in and to load source from,
to avoid interference, to simplify cleaning, and to
take 'in_testdir(.)' into account?).
Yes, something like that.
Cheers,
Simon
_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc
