stability

[Derek Parnell]

On Sun, 24 Feb 2008 12:24:22 -0800, Walter Bright wrote:

> Derek Parnell wrote:
>> I refuse to believe that Walter could not create a better D compiler if it
>> was known what a D compiler was supposed to do. 
> 
> This is so true. Back in college, I attended a lecture about the 
> research process. A researcher would tend to erratically go all over the 
> place trying to find an answer. Once he found it, he'd write up a paper 
> explaining his process, skipping all the dead ends, reversals, and 
> mistakes to make it appear as a straightforward, logical progression 
> from problem to solution.
> 
> I don't think it is possible to design a useful, non-trivial language 
> without evolutionary feedback.

This is also very true. A design process is always enhanced by
'prototyping' in order to uncover what the real requirements are. The other
activity which is just as important in this process, and most others too,
is serious and fearless peer review.

It might be said that we have in DMD, the prototyping activity. With the
forum and Walter's closer associates, we could be seeing an informal
variety of peer review. The DigitalMars documentation for D could be argued
as being the current version of the specification. 

Now all we need to see is the specification completed, in the sense of
filling in the gaps from what's in Walter's head and cleaning up the known
issues with it. Then we can continue the cycle by bringing the compilers
"up to spec", which will feedback into an improved spec, and around we go
again. Incremental improvement of D, the compilers, the libraries, and the
D community's sanity. Well that last one might be a stretch goal ;-)

-- 
Derek Parnell
Melbourne, Australia
skype: derek.j.parnell