[mercury-users] pprint really does have a performance bug

[Ralph Becket]

I forgot about Fergus' report of serious performance problems with
the new pretty printer library (pprint.m) and just got bitten by
it myself.

I've been doing some performance tests on plan graph generation
which, up to a limit, should have combinatorial runtime costs.  I
was very surprised to observe super-combinatorial results, even on
very small data sets (no results after 16+ hours on n = 8).  A little
more digging reveals that I do get the right results from the
algorithm I'm testing, but that pretty printing is what's costing a
packet!

You have been warned!  pprint is not yet suitable for printing large
terms.  As I said on mdev, large term printing will have to wait until
either
(1) I (or some kind soul) sort out the performance bug or
(2) Tyson gets the laziness transformation working before (1) happens!

This is what you get for transliterating Haskell programs and hoping 
for the best :)

Ralph

--
Ralph Becket      |      MSR Cambridge      |      rbeck at microsoft.com 
--------------------------------------------------------------------------
mercury-users mailing list
post:  mercury-users at cs.mu.oz.au
administrative address: owner-mercury-users at cs.mu.oz.au
unsubscribe: Address: mercury-users-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-users-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------