While creating a list that would add support for Xtrema's
objects, I discovered a couple of things that may be of
interest to this
Note that the xList objects used in the tests are binary-wise
lists, that are accessed by a different engine.
The fist test was string only. Searching for a string within
a list of
strings. xList seems to be twice as fast as director's list,
which was not
really a surprise, since the xstrings that are used
internally for the
comparison, are much faster than director's strings.
What was a surprise however, was searching for a string, in a
symbols, or other non-string values.
In such a case, director will perform string comparison in
all values - the
xList objects test if the types match, according to my tests
at least, were
150 times faster.
Searching for a symbol in a list of symbols was again faster,
but by a
'normal' 70% - at least on my test machines.
Now, I wonder.. Is such a performance loss justified?
Meaning, do you really
want director to return the position of "myData" when you
#myData? Cause there seems to be a huge price for that, at
using lists containing various data types.
The archive contains the movie + a projector + a build of the
includes a first version of the xList object - it supports
commands: append, getPos and getPosI.
Case anyone tries the movie, or tests the xlist object with
types, I'd be interested to see the results from various
machines (you can
send them to the mail you'll find at the first page of the