Re: [Gnumed-devel] audit trail question

[Karsten Hilbert]

> 1 keeping backups of entire rows upon change
> 2 keeping backups of the changed columns only
> 3 logging the queries leading to said changes
> 
> 1 allows for much easier access to a revisional history of the
>   records, it can be implemented with procedural languages
>   alone but it always keeps the entire column, this is
>   available, needs an audit table per audited table
> 
> 2 access to a revisional history of the record is much more
>   involved, this is often faster, it does not store as much
>   redundant data as 1 but it does require a backend loadable
>   module (in C, usually) to do so, this is also available, can
>   audit all data to one audit table

We seem to be leaning towards 1) or 2). The current
implementation in server/utils/audittrail.py is written for
1). I have not seen any discussion so far on the relative
benefits of 1) and 2). I'd like to see more arguments either
way before committing to any of the two.

Karsten
-- 
GPG key ID E4071346 @ wwwkeys.pgp.net
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346