Problems dealing with GHC incompatibilities

[George Russell]

I used to have a lot of trouble with UniForM / HTk every time there was a new GHC release,
because we use a LOT of GHC extensions.  Sometimes a GHC release wouldn't work at all and
I had to submit a shopping list of bugs to the GHC Team, in the hope that they would issue
a version X+0.01 which we could use.

However I'm glad to say I had virtually no trouble with the upgrade to GHC5.04.  I think this
is partly because I tried out ghc5.03 several times before ghc5.04 was released, partly from
the snapshot releases, and partly from the CVS version.  I made sure the UniForM code worked
with ghc5.03.  Where I had real trouble I was able to complain loudly to the GHC team, and get
them to fix the problem, in time for ghc5.04.  For example, I think I had two problems with
ghc-pkg, which Simon M was kind enough to fix.

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).
(2) If you are responsible for a lot of Glasgow Haskell code, you make sure it works not only
with the latest official GHC release, but also try to keep it in step with the snapshot releases.
We use the -cpp option and combinations like

#if (__GLASGOW_HASKELL__ >= 503)
-- new stuff
#else
-- old stuff
#endif

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.