assert semantic change proposal

[via Digitalmars-d]

On Thursday, 7 August 2014 at 09:26:01 UTC, Walter Bright wrote:
> On 8/7/2014 2:10 AM, "Marc Schütz" <schuetzm at gmx.net>" wrote:
>> On Wednesday, 6 August 2014 at 21:57:50 UTC, Walter Bright 
>> wrote:
>>> On 8/6/2014 10:18 AM, Sean Kelly wrote:
>>>> So from a functional perspective, assuming this is a 
>>>> multithreaded app, what
>>>> should the procedure be when an AssertError is thrown in 
>>>> some thread?
>>>
>>> Print message, stop the process and all its threads.
>>>
>>> Rationale: the program is in an unknown, invalid state, which 
>>> will include any
>>> thread (or process) that has access to memory shared with 
>>> that failed thread.
>>
>> Strictly speaking we would need to kill only those threads.
>
> No, all the threads. Threads share memory, so corruption can 
> spread from one to the next.

Yes, that's what I was saying: "those threads ... that ha[ve] 
access to memory shared with that failed thread".

But even without real sharing of memory, a failed assertion in 
one thread might result from invalid data being sent to that 
thread from another one (via spawn or send). Better to just kill 
all threads.