Subj : Re: ECMAScript standards committee
To   : Shanti Rao
From : Brendan Eich
Date : Mon Oct 11 2004 08:55 pm

Shanti Rao wrote:
> Georg Maaß wrote:
> 
>> Why should JavaScript 2.0 less stable than JavaScript 1.5?
> 
> 
> Did you notice now long JS1.5 was under development? JSVERSION_1_5 got 
> added in Sept 1999. Five years later, there are still a few bugs.


There are always bugs in any non-trivial, evolving codebase.  If you 
think JScript, or KJS, or NOMBAS (now OpenWave) is bug-free, I have a 
bridge to sell you.  The ECMA-262 Edition 3 spec itself has way too many 
bugs (see http://www.mozilla.org/js/language/E262-3-errata.html for a 
partial list that may itself have bugs!).

It's naive to complain about the presence of "a few bugs" in something 
that's under active development, and of the complexity of SpiderMonkey. 
  Complain about bugs, by all means -- but do not expect zero bugs on 
all platforms, for all use-cases, until well after the code and the 
language are dead.


> I 
> concede that a complete redesign might help, but I'm not the expert on 
> that.


Complete redesign of the language is not under discussion.  A complete 
engine redesign could happen, but many parts would work the same as 
existing implementations, either SpiderMonkey, Rhino, or Narcissus. 
There's no point in redoing an operator precedence parser that does the 
job right already, e.g.  Another example: the Mono "JScript" 
implementation is basically porting Rhino from Java to Mono C# and the 
Mono CLI implementation, from what I can tell.

The main issues with SpiderMonkey, judging from bugs filed against it, 
are memory problems to-do with how it schedules GC (or fails to schedule 
GC), and how it allocates objects.  I'm working on one fix to memory 
behavior that doesn't require complete redesign of anything -- see 
https://bugzilla.mozilla.org/show_bug.cgi?id=157334.

/be

.