Yeah, I ended up being a little vague after writing a really long one trying not to cloud it. Erg.
Doing multiple records/rows and then using the most recent did occur to me but I figured there might be a lower memory way or A Way It Is Done(tm). My conception of how I would do it if I could get it was:
create table changes ( id char(10), # DB's other tables all have uniquely prefixed ids diff mediumtext, timestamp datestamp );
And then stringify + concat all the new values of the cols, diff it against the old... It would need to do diffs and patches on strings. Text::Diff looks helpful for the first part at least. Patching (reverting) doesn't seem as obvious.
Now that this problem is broken down against the simplicity of the problem (as DaveH and Corion mention) I think repeated records would be easier to implement less and prone to problems over all. Thanks!
In reply to Vagueness and patched strings
by Your Mother
in thread Revision control in relational databases
by Your Mother
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |