Terms - mirrors & copies

[Martin Hill]

Roy Williams wrote:

>> mirror
>> copy
> 
> 
> I have learned over the last 2 years to be very afraid of these. 
> 
> Two things are the same ONLY WITHIN A SPECIFIED CONTEXT.
> 
Hmmm... I would have thought things are either the same or they are not. 
  Where they are not they might be _related_

So in describing which term is which, how about:

A 'mirror' is a deliberate attempt to keep things the same. If the data 
is static this is easy (ignoring left and right swaps in this analogy 
:-). It may be there are very rare such true mirrors, but we have one at 
the ROE for failover, and I'm sure we're not the only ones!

Where the data is changing (eg, surveys still being built and propogated 
to other remote sites at intervals) describing a 'mirror' will be more 
tricky and we should I agree treat that particular case with caution. 
Probably best to treat it as a different but related dataset/collection.

A 'copy' from the original is a 'branch' on the original.  Both original 
and copy are free to change.  The description of the 'copy' will be a 
different resource document that might also start as a copy of the 
original resource document.  We *might* include in their descriptions 
some kind of change history, and we *should* include links to variations 
on datasets ("If you like the ROE 2MASS dataset with its whatevers, you 
might also like the OpenSkyQuery 2MASS dataset with its whatevers").


-- 
Martin Hill
www.mchill.net
07901 55 24 66
0131 668 8100 (ROE)