On 4 Jan 2018, at 21:13 (-0500), @lbutlr wrote:
On 4 Jan 2018, at 11:47, Bill Cole
<sausers-0150...@billmail.scconsult.com> wrote:
On 3 Jan 2018, at 15:42, @lbutlr wrote:
There is no requirement that the right side be globally unique, just
that the entire message ID is globally unique.
Right. And any software that can use localhost (or any other
unqualified name whose meaning is contextually variable) as the right
hand side is likely to be doing so on multiple machines that don't
know about each other and so generally cannot know that they are not
generating duplicate MIDs.
Sure, but depending on how the MID is generated it can certainly be
statistically unique. As I said earlier, it only takes 256 bits to get
an ID within spitting distance of the number of atoms in the universe.
Should be unique enough.
Not even that. A standard UUID has 122 bits of entropy. To *probably*
have *one* collision in that space, you'd need to generate 1 billion
UUIDs per second for about 85 years. That should be good enough for
naming unique things including email messages for as long as any one
person cares, but if you want it more solid you could put a UUID of one
of the node-specific versions on one side of the @ and a random UUID on
the other: 244 bits, won't collide in any space-time region visible to
one observer.
--
Bill Cole
b...@scconsult.com or billc...@apache.org
(AKA @grumpybozo and many *@billmail.scconsult.com addresses)
Currently Seeking Steady Work: https://linkedin.com/in/billcole
