[LINK] Looking for some advice from the link 'brain'

[Adrian Chadd]

On Mon, Apr 02, 2007, Alan L Tyree wrote:

> Does it matter that OO now saves data according to an ISO standard?
> Unless the specifications of the standard disappear, isn't it possible
> to recover? It no longer depends upon the existence of OO or any other
> particular bit of software.

The process of copy and reencode has two properties:

* the data representation is updated to match whatever you're now using, and
* the physical data representation/encoding on some media is updated to match
  whatever you're now using.

I have a pair of SCSI DDS-2 drives sitting around at home purely to loan to people
who have "mission critical urgent data that needs to be read now but we threw away
our SGI Indy/Sparc Ultra1 and DDS-2 drive years ago" related issues.

The data representation format was simple - 7-bit clean ASCII.

The data archival format is simple - tar.

The data encoding format wasn't so simple - no DDS-2 drives.

It gets worse if you're wanting to read some of the really old stuff (for values
of old greater than my age but less than, say, 100, as early storage media
were analog/mechanical and easier to build readers for) - say, the data tapes
storing recorded telemetry and data from one of the early NASA space probes.
I think it was Pioneer. If I remember right:

* They found the media
* No documentation on the data format was initially found
* All the people who knew the data format were dead
* All the readers were busted.

I forget how this was eventually worked around.

(I could get all "inherent data structure" on you and start spitting out references
 to papers on distilling structure from unstructured data - text mining, genetic
 analysis, language analysis, etc, but (a) its tangentally related, and (b)
 would be some of the modern techniques to use to figure out arbitrarily, non-encrypted
 data from the past :)




Adrian