[llvm-dev] The undef story

[Peter Lawrence via llvm-dev]

Tim,
       Can you make a case for why this isn’t the right thing to do ?

I think this is a no-brainer because this is the situation we would be
in if folks hadn’t gone off the deep end with “poison” which hasn’t
actually solved any problems (other than the mis-adventure with
“+nsw” which never actually needed “poison” in the first place).

And AFAICT this is the situation that gcc is in, it gets along just fine
without “poison”.

If you think there is a problem that this approach doesn’t solve then
you need to demonstrate it.

Peter.


> On Jun 28, 2017, at 4:01 PM, Tim Northover <t.p.northover at gmail.com> wrote:
> 
> On 28 June 2017 at 15:33, Peter Lawrence via llvm-dev
> <llvm-dev at lists.llvm.org> wrote:
>> The correct software engineering decision here is to fix the definition of
>> “undef”, delete “poison”, and not hoist “nsw” attributes.  That is a no-brainer.
>> There is nothing to try out, or test, ormeasure. That is simply the way it has to be to avoid
>> the current  set of problems.
> 
> Declaring victory and expecting everyone in a foreign community to
> fall into line is probably the the second-least productive approach
> that could possibly be taken in this situation. Surpassed only by
> actually insulting a regular at the same time.
> 
> Tim.