CLIM mail archive
[Prev][Next][Index][Thread]

Re: Moving objects and incremental redisplay


Here are my 2 cents on this issue.

1.  There is at least one bug in coordinate-sorted-set output histories
    that cause output records to get "lost" via the sorting algorithm.
    This bug can be found by running scigraph and deleting the line
		:OUTPUT-RECORD (MAKE-INSTANCE 'CLIM:R-TREE-OUTPUT-HISTORY)
    from scigraph/frame.lisp.  It can be seen by trying to drag a graph
    annotation.  When you "let go", the most common problem I have heard
    is (a) the annotation loses its sensitivity and (b) replay of output
    records starts losing.

    This bug can cause incremental redisplay to lose, though it is not
    a bug in incremental redisplay per se.

2.  Beyond #1, most of the problems I have had with incremental redisplay
    since clim 2.0 beta have turned out to be bugs in my own code.  This
    is based on quite a variety of experiences.  It is certainly more reliable
    than clim 1.1 and dynamic windows.

3.  Nevertheless, proper care and feeding of incremental redisplay
    caching points is VERY HARD.  It takes a good model of what incremental
    redisplay does, and even so it is hard to debug because clim takes
    over completely once all the output records have been generated once.
    There is clearly a need for lots of documentation, rules of thumb,
    clim lore, good examples, and prayer here.

4.  For simple problems, proper incremental redisplay can take as long
    to set up as "rolling your own".  Therefore I think its value is
    for problems that are not simple, such as graphs (the hierarchical kind).

jeff morrill
Follow-Ups: References:
Main Index | Thread Index