jastrachan | 18 Feb 10:25 2005

Re: Re: static versus dynamic typing policy

On 15 Feb 2005, at 23:46, Mike Spille wrote:
> At 01:37 PM 2/15/2005, you wrote:
>> On 15 Feb 2005, at 18:25, Chris Poirier wrote:
>> I'm not ignoring his points - they are very valid concerns -  I was 
>> trying (and obviously failing miserably) to say, lets keep debating 
>> this issue and make the right decision and not use the current status 
>> of the Antlr grammar or Groovy AST mappings as excuses for stopping 
>> the discussion.
> I'm not attempting to stop the discussion.  I'm distinguishing between 
> what can be implemented in a timely fashion and what's questionable.
> A case in point, the latest TCK test reports:
>         Tests   Failures        Errors  Success rate    Time
>         361                2     218           39.06%     527.610

FWIW there were some bugs in the test case itself. We're now close to 
60%. Mostly constructors and array initialisers are breaking the 

> It's now mid Feb and the tests haven't reached 50% yet.  I've heard 
> the arguments about open source being different from closed source 
> projects - but on a closed source project this would be a sign of 
> grave concern.
> In this light I think discussing what's implementable is directly on 
> point.  Academic ideas sound good in theory - but not so good when 60% 
> of your tests fail after months of effort.
> To understand what I'm talking about James, go back to the Bileblog 
> comments from last December.  I don't mean this to bring up a sore 
> point to show other people's point of view.  Look at what was said 
> back then.  Now take a cold, hard look at what can be seen in 
> February.  Imagine someone reading those comments in early December 
> and then forgetting Groovy for a couple of months and coming back at 
> it cold today.  By just reading your comments I think that person 
> would be in for a big shock.  And a big reason for this is because it 
> seems that "implementability" isn't really high on the priority list.

Hmm; we were discussing a new grammar & parser in November, work 
started in December - then after holidays & festive period integration 
work started January and by end of Feb we should pretty much have an EA 
version of the new grammar fully integrated into the runtime passing 
all of the old classic test cases. That's not bad progress considering 
the size of the job.

Also remember how long other good languages take to get right (python, 
ruby, java, C#) took many years.