The UserClass runtime was able to sort out the init args that had
setters and those that did not, and it would install the ones that
didn't have setters in the prototype (hence they were guaranteed to
all be defined before any setters ran, in fact, they are all defined
before any method runs, including the construct method). This may be
the source of the difference that Henry was seeing in the grid demo.
Presently, I can't do this at compile time, because the class model
for built-ins is incomplete (basically empty). I don't like to do it
this way, but I am thinking that I will have to just hard-code the
list of setters that exist in the built-in classes for now.
If Mr. Doc Tools has a spare cycle, there might be a way we could use
a variation on the doc tool to extract setters from the LFC code and
spit them out as a .xml or .properties file that the compiler can load.
- [Laszlo-dev] One difference with the class-compiled user c... P T Withington
-