Re: msg-id

[John Stanley <stanley@xxxxxxxx>]

Thorfinn <thorfinn@xxxxxxxxxxxxxx>:

>Go back and read again.  Particularly regarding field order in the two
>algorithms.

What "two" algorithms? I proposed only one. 

> He doesn't use an algorithm which duplicates message ids. However, when
>compared to *your* algorithm, at some point, it is possible that your
>"no duplicates" algorithm will have generated IDs which happen to be
>identical to his "no duplicates" algorithm.

So, my algorithm produces duplicates because at some time in the future 
someone else's algorithm might duplicate something I've already produced? 
No, sorry. The duplicates are not produced by my algorithm. They come from 
a different algorithm. If that algorithm produces duplicates, then that 
algorithm is broken.

>Both, considered on their own, will generate unique IDs. 

Actually, not they don't both produce unique IDs. You've just said that
the second algorithm might produce an ID that has already been produced by
mine. Thus the second algorithm produces duplicates. But that's
irrelevant, since the discussion is about whether the algorithm I
suggested would do so. And it does not.

>So, where is your uniqueness?  Non existent, unless you can
>guarantee that you will own that domain name forever and never give it
>up.

Get a clue. I cannot guarantee uniqueness in the actions of someone else, 
if they choose to use an algorithm that can produce duplicates. Their 
choice to use such an algorithm is a) not under my control and b) not a 
failure in my algorithm. I CAN guarantee uniqueness in the algorithm I 
use.

>> Yep. Got that. Unfortunately, "MUST" and "unique" combine to zero. Either 
>> you don't mean "unique" or you don't mean "MUST". Which is it?

>We mean MUST, and we also accept that MUST applies in the universe of
>practicalities, where nothing is actually of zero probability, so "close
>enough" is good enough.

Until you show me an operating system that produces the same process id 
for two different processes, then the algorithm I presented has zero 
chance of producing duplicates. Even then, that same operating system 
would have to provide no means of locking such that unique sequence 
numbers could be produced, and then do all of this within one second.

Zero is better than "good enough", when zero is the goal. If anyone 
doesn't agree with that, please speak up.

>> >They are DIFFERENT message-ids. Yes?
>> Not as far as I can tell.

>Well, now that you've actually changed what Charles wrote, they look the
>same.  

Changing what someone said is a pretty serious claim, pal. You better have 
facts to back that up. I changed nothing. I cut and paste exactly what I 
saw. The two look identical to me. Do they look different to you?

><12345678@xxxxxxxxxxxxxxx>
><12345678@XXXXXXXXXXXXXXX>

>are different Message-IDs, though.

Yep. I've not said otherwise. In fact, I think I was one of the people who
was convincing Charles et.al. that the RHS is case sensitive. Did you 
think otherwise?

"Charles Lindsey" <chl@xxxxxxxxxxxxxxxx>:

>Consider two computers on the same LAN, mounting files from a common
>source by NFS. They might well be using the same pid at the same instant
>(once in a thousand years, maybe).

Yep. And if they are on the same lan, I will have given them different 
names, so the RHS will be different. On one system, one must guarantee 
that the LHS is unique; between systems, the RHS takes care of it.