Problems dealing with GHC incompatibilities

George Russell ger@tzi.de
Fri, 26 Jul 2002 15:26:21 +0200


Simon Marlow wrote:
> 
> > So my suggestion is that
> > (1) The GHC team continue to issue snapshot releases as often
> > as possible (I don't know why they
> > don't make the automatic builds generate them every night).
> 
> We could, and in fact the nightly build system has support for
> publishing the resulting source & binary bundles.  One consideration,
> though, is that someone pulling down a snapshot will want a clue as to
> whether that snapshot is likely to be not completely broken, and without
> reading cvs-ghc@haskell.org.  So we tend to pick snapshots at moments
> when we know things are in a reasonably stable state.

What version of GHC do you use at Microsoft?  Do you have a last-not-broken version
which you use for developing GHC or what?  If there is a last-not-broken version it would
be good to have access to it.

For quite a long time I ran an automated system which compiled GHC every night from
CVS (using the previous version) and installed it on our system if the compile completed.  This
actually worked quite well, though I stopped doing it because the compilation system got
more complicated, I had problems doing it on Sparc/Solaris, and it was too much trouble.
If the compile finished it was unusual for the compiler itself to be hopelessly broken;
to be on the safe side you could check it was able to compile Happy and Haddock.  Something
like this might be good.  Of course it would have to be made very clear that such snapshot
releases were for evaluation only, and that only the official releases were to be relied on.

> 
> > By the way, Simon M, how long can we rely on the hslibs
> > library structure?  Because some time
> > I shall have to schedule time to move over to the new
> > structure, and I'm not looking forward to
> > it.
> 
> Perhaps they'll go away in 6.00... how do people feel about that?  The
> 5.04 release will continue to be supported up until the release of 6.00,
> which is likely to be at least 6 months (probably longer).
6 months is, from my point of view, fine.

George