[Zope] The "undo" posibility and the size of data.fs

[Toby Dickenson]

On Monday 23 December 2002 12:22 pm, borge@tsar.no wrote:

> We now do wonder if zope's "undo" possibility is an
> explanation to this large increase in data.fs; this
> since zope holds the states inbetween the transactions
> (four transactions for each question).

(3k per question)/(4 transactions per question) =3D 800 bytes per transac=
tion.

This is normal if each transaction modifies your emloyee object, and each=
=20
employee is roughly 800 bytes in size.

You could use sessions to combine this into one transaction - storing=20
intermediate changes in a volatile session object. This would give you a=20
factor of 4 space saving.

> If it is this
> undo possibility that causes the large increase in
> data.fs

Thats just the way FileStorage works. Yes, it is used to implement undo.

> we would like to know if it is possible to
> get around this "undo" tracking af zope.

There are some storages that dont support undo, but nothing that is=20
production-ready :-(

Your best option is to pack regularly using cron.


--=20
Toby Dickenson
http://www.geminidataloggers.com/people/tdickenson