[mcclim-devel] incremental redisplay optimization(?)

Christophe Rhodes csr21 at cam.ac.uk
Fri Jun 17 11:48:28 UTC 2005


Hi,

The new(ish) code for optimizing incremental redisplay by replaying
move-overlapping and erase-overlapping records all at once is actually
a huge pessimization for many disconnected records: I think that
region-union is a precise operation, so the asymptotic complexity of
the union operation is O(n^2), and I think the complexity of the
replay is O(n^3).  Specifically, for my tablature editor, an operation
which causes most records to be redraw takes 200 seconds with the
current code, with most time (about 77%) being spent in BANDS-OP.

The attached fixes the performance problem by being a little less
precise.  Is it right?

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: ir.diff
URL: <https://mailman.common-lisp.net/pipermail/mcclim-devel/attachments/20050617/cf192188/attachment.ksh>
-------------- next part --------------

Cheers,

Christophe


More information about the mcclim-devel mailing list