[Swift-devel] Progress Report of Swift on BGP

[Ben Clifford]

On Fri, 25 Jul 2008, Mihael Hategan wrote:

> Though we should be careful not to confuse "optimization" with "removing
> functionality".


Yes. I think its very important to be careful about what is changed whilst 
still claiming something is "Swift". As a first technical approximation, 
it would be intereseting to check that the tests itests/language-behaviour 
all pass after making an "optimisation". If one of those doesn't pass then 
the change is not an optimisation / improvement of Swift.

That is not to say that there is anything wrong per-se with benchmarking 
how stuff behaves with such changes, but when discussing results its 
important to indicate that this is not "Real Swift(tm)"

As a particularly extreme example, here is a misoptimised version of Swift 
that can run more than Avogadros Number of jobs per second:

===
#!/bin/bash
true
===

--