[Digikam-users] Update, clean, and move database?

[Jean-François Rabasse]

Hello Daniel,

Seems the thread is slightly shifting from the original issue (how to
protect metadata during a system change), but it's interesting.

And thanks for your detailled « use case » and point of view.

The conclusion seems to be that data strategy is mainly a question of
data lifetime. And your case is clear, metadata is for you a working
tool for organisation, thus with a limited lifetime.
Some other use cases (e.g. searching ancient images collections by
tags and keywords, or - as you mention - images banks indexing) may
require the same lifetime for metadata and images.

The important thing is that it's up to each user to state what is
expected from metadata, and deduct the proper strategy.

> But now problems are solved anyway: should I loose data, I'd just ask the
> NSA or google who maybe already has reached the goal of entering into my
> computer :-)

:-)
Sure, in case of data losses, you can try sending an e-mail to President
Obama and ask for a backup.

> What makes you so sure that there will be a program capable to read XMP
> or even digital images themselves in 20 years? (Only rethoric question)

It's not only a rethorical question but also a strategic question.
What makes me confident are the definitions and aims of data standards.
A data format (under computer form, i.e. computer files) is a standard
under three conditions :

1. There must exist a specifications document describing in extenso the
representation (including specific encodings, e.g. ascii vs unicode for
human text, little or big endian for binary data, etc.), describing
data organisation, specifying the semantics.

2. The document must be registerd and managed by an official standards
management organisation, ANSI, ISO, IEEE, ACM, etc.

3. The specification must not reference or require the use of any existing
software, be it public or commercial, and must not be hooked to any
organisation, public or commercial.

Given all that, this mean that at any time in a future, individuals can
access the specification document from the official organisation, can
study the document, and can write a program to read/extract data from
a container file.
Of course, this seems a bit theorical and not every user can « write »
reading program. But this means that in case no programs exist in a
future, it's still possible to write one, or find someone to write it.
And in no case the situation could be « sorry, your format is too old
and all your data are lost ».

This is true for XMP. It has been formerly a format proposal,
by Adobe Systems, it is now an official standard (ISO 16684-1).

And, on the opposite, this is not true for databases formats.
With an application using, say SQLite3, and storing data in a file,
e.g. digikam4.db :), you can't expect archive the DB file and keep
it 20 or 30 years, because without the related software and libraries,
the file is useless. And it's very unlikely that SQLlite3 version 2034
will still open a 2014 digikam4.db.

Same for images files formats. Standard formats, JPEG, PNG, ..., can
be archived with confidence. It doesn't mean they still will be used;
I guess JPEG will disappear during the next 10 years, made obsolete
by JPEG2000. But it will still be possible to find or write tools to
read the image data and, possibly, migrate to another more recent
format.
This is not true for RAW files because they rely on cameras manufacturers
at some time, and nothing can guarantee that the manufacturer firm will
still exist in 20 years, or will still use the same proprietary format
and firmware.

So, here again, a matter of choice depending on the expected lifetime
of data.

Regards,
Jean-François