>>>> "NP" == Nick Pisarro
<nickp(a)aperture.com> writes:
NP> Storing diffs in the 'old' table. [...]
NP> Having written such code in the past, I plan to create
NP> a write up exploring this idea. Has this idea been discussed
NP> amongst the developers? What are the gotcha's?
I brought it up on wikitech-l a few weeks ago:
http://mail.wikipedia.org/pipermail/wikitech-l/2004-February/008330.html
Once again illustrating the lameness of mailing lists for design
discussions. I've copied that email to meta:
http://meta.wikipedia.org/wiki/Reverse_diff_version_control
...as a first step in really hashing out this idea.
I think the main issues were the following:
* Any decrease in performance -- even for relatively rare actions,
like retrieving a historic page version -- is really not OK.
* It'd be great to reduce database size: saves wear-and-tear on
storage devices, saves time on e.g. backups, etc., etc.
Anyways, it's probably a good idea to follow up on. Let's work on meta
to whip it into a good proposal.
~ESP
--
Evan Prodromou <evan(a)wikitravel.org>
Wikitravel -
http://www.wikitravel.org/
The free, complete, up-to-date and reliable world-wide travel guide