[mercury-users] 234-Trees vs alternatives in the compiler

[Fergus Henderson]

On 16-Dec-1998, Peter Schachte <pets at cs.mu.OZ.AU> wrote:
> On Wed, Dec 16, 1998 at 08:31:06AM +1100, Andrew Bromage wrote:
> > The 234-tree implementation has changed somewhat since these experiments
> > 
> > It would be interesting to try the experiments again with one or two
> > more exotic data structures.
> 
> How about backtrackable expandable hash tables, using trailing

That would be nice, but it would have a different interface.
In particular, that would need to use mdi and muo
("mostly unique input"/"mostly unique output") modes
instead of in and out modes.  So, to use this, we'd need
to modify our code to use the new interface.

Actually most of our maps use abstract keys like pred_ids,
proc_ids, or vars, so if/when we modify the code so the
variables tables are singly-threaded through the code,
it would be better to use `store' for them rather than
using a hash table.

> and user-defined equality?

That's a separate issue, really.

-- 
Fergus Henderson <fjh at cs.mu.oz.au>  |  "Binaries may die
WWW: <http://www.cs.mu.oz.au/~fjh>  |   but source code lives forever"
PGP: finger fjh at 128.250.37.3        |     -- leaked Microsoft memo.