[Manifold-l] download problems

[Dimitri Rotow]

> download large files successfully. Hence my suspicion that 
> the setup of the manifold server is at least partially to blame.
> 
>

and 

>     Don't doubt it at all.  Two different machines; 6 hours apart; 
> same result.  I'll wait for the real thing.  Linus
> 

I wouldn't be in a hurry to ignore broad information in favor of very thin
samples.  That's just a formula to overweight your own experience and ignore
what is really going on. The data are that there are vast numbers of
downloads of updates.  Experienced Internet people know there are many
reasons why downloads, especially those going half-way around the planet,
might have problems in transit.  Combine the two and it is virtually
guaranteed that there will be some failure rate, say, 1 in 500 or higher,
even with perfect servers.

It's similar, I suppose, to how some folks leap at considering every
unexpected result a "bug," when in fact as a matter of statistics it would
be more efficient to invest into a careful examination of what is being
done.  In this case as a matter of simple statistics you would be better
served by accepting the truth that virtually everyone else but you is having
successful downloads and considering, carefully, that the problem is not in
the server but more likely in the chaotic chain of Internet connections
between Manifold's server and your desktop. 

Consider the "same result" comment: Same Internet link by any chance? Same
ISP? Same proxy server? Some people will rig up gateways or proxy servers to
cache transfers.  So if you hit the same URL it returns the same content
(the idea is to save bandwidth by not re-fetching something that's already
been fetched).  That's fine unless the transmission was damaged before it
got to the cache, in which case you simply get the damaged cache.  That will
happen to anything downstream of the cache regardless of when or where the
URL is hit.  

No doubt people more familiar with Internet innards can suggest other
scenarios. The point is that there is much that can go wrong even when
servers function perfectly.  Can servers misbehave?  Sure.  But given the
statistics that's not the first place to look until the data indicate it's
time.

In the meantime, I assure you that what I wrote in my previous posting is
indeed correct, that very many other people have downloaded exactly that
same file without difficulty.  

Cheers,

Dimitri