|
trivial random ports for deterministic seeding
Peter McGoron
(31 May 2026 16:58 UTC)
|
||
|
Re: trivial random ports for deterministic seeding
Shiro Kawai
(31 May 2026 18:12 UTC)
|
||
|
Re: trivial random ports for deterministic seeding
Vincent Manis (he/him)
(31 May 2026 18:30 UTC)
|
||
|
Re: trivial random ports for deterministic seeding
Wolfgang Corcoran-Mathe
(01 Jun 2026 00:11 UTC)
|
||
|
(missing)
|
||
|
Re: trivial random ports for deterministic seeding Wolfgang Corcoran-Mathe (01 Jun 2026 02:50 UTC)
|
||
|
Re: trivial random ports for deterministic seeding
John Cowan
(01 Jun 2026 06:25 UTC)
|
||
|
Re: trivial random ports for deterministic seeding
Peter McGoron
(01 Jun 2026 11:55 UTC)
|
||
|
Re: trivial random ports for deterministic seeding
Wolfgang Corcoran-Mathe
(02 Jun 2026 00:29 UTC)
|
||
|
Re: trivial random ports for deterministic seeding
Wolfgang Corcoran-Mathe
(01 Jun 2026 00:14 UTC)
|
||
|
Re: trivial random ports for deterministic seeding
Peter McGoron
(01 Jun 2026 00:39 UTC)
|
||
|
Re: trivial random ports for deterministic seeding
Peter McGoron
(01 Jun 2026 00:52 UTC)
|
||
(CCing John's off-list reply.) On 2026-05-31 21:12 -0400, John Cowan wrote: > > "All zeros" is definitely a > > problem and must not be used. > > I'd say "must signal an error". I'm not so sure about this. Would this mean than an implementation must signal an error if it reads "enough" bytes and they all turn out to be zero? To me it seems preferable in this case to retry, on the assumption that the port's output is *not* always zero. But then how many times do you try to initialize before signaling an error? I've added that implementatins MAY allow initializing with an all-zeros port, but I'm open to arguments for different behavior. I just think "must signal an error" is a bit hard to interpret in this case. > Perhaps we should likewise set a minimum size, like 8 bytes. That is, an *initializer* port must produce at least 8 bytes or an error is signaled? -- Wolfgang Corcoran-Mathe <xxxxxx@sigwinch.xyz>