[Dlang-internal] Regression control & breaking changes policies

[Walter Bright via Dlang-internal]

On 12/6/2016 5:56 PM, Dicebot wrote:
> THERE IS NO SUCH THING AS MINOR REGRESSSION

BTW, my uncle was a research chemist for a pharmaceutical company in 
Switzerland. He told me long ago that there was no such thing as a drug without 
harmful side effects. All drugs have a "therapeutic index", which is the ratio 
of the good they do vs the harm. The researcher's job was to make that index as 
high as he could.

It's the same with any real world product. Nothing could be built if, for 
example, safety is a top priority that trumps all else. Of course, companies and 
governments always say that safety is their top priority, but that's because 
they're forced to by the media who will excoriate them for saying anything else.

We do need processes and rules, but they should serve as guides for good 
judgement, not as substitutes for good judgement.

And that's where we need your help and judgement - in deciding what that 
"therapeutic index" is for the PRs on a case by case basis.