We should get rid of it as it never fully worked and has surely rotted
badly by now.
At one time, profiling by Adam showed that it was expensive to have
too many levels of class derivation. But, once we are compiling to
JS2 classes, if we were going to do anything like this, we should be
doing it in the script compiler, leveraging of the mixin expander,
rather than doing it only for LZX classes.
On 2008-02-08, at 18:15 EST, Henry Minsky wrote:
There's still code in the tag compiler for inlining of LZX classes. I
don't know if it works, but am guessing probably not.
It seems to require a special <?lzc ..> processing instruction which I
don't recall ever seeing used.
Anyway, is there any point in trying to preserve it, or should we
get rid of it when we get a chance? What was the thinking about how it
was supposed to improve
performance?
--
Henry Minsky
Software Architect
[EMAIL PROTECTED]
