Re: rand is not ISO C compliant in Cygwin

["Allen, Norton T." <allen@huarp.harvard.edu>]

On 11/11/2023 1:25 PM, René Berber via Cygwin wrote:
> On 11/11/2023 10:50 AM, Allen, Norton T. via Cygwin wrote:
>
> [snip]
>>> The srand function is not required to avoid data races with other
>>>     calls to pseudo-random sequence generation functions. ..." 
>> That is not the same as "... required never to avoid data races ...". 
>> "not required" means the sentence is not specifying--not 
>> requiring--any behavior, so you should not depend on the described 
>> behaviors.
> [snip]
>
> The elided part on Bruno's message is:
>
> "The implementation shall behave as if no library function calls the
> rand function."
>
> Which is the point Bruno is making with the sample code.

I would still assert that if the implementation is doing what I 
suggested (maintaining independent state for each thread) it would still 
meet that criteria, although I admit it is debatable. That said, the man 
page says:

"rand and srand are unsafe for multi-threaded applications. rand_r is 
thread-safe and should be used instead."

I read that as "all bets are off" if you are using them in a 
multi-threaded application.