Hi David,
feel free to experiment, however you might want to have a look at our
take to improve version handling in magnolia:
http://wiki.magnolia.info/display/DEV/Concept+mixin+versionable
http://jira.magnolia.info/browse/MAGNOLIA-2086
http://philipp-bracher-magnolia.blogspot.com/2008/05/magnolia-36.html

Reducing usage of mix:versionable has performance implications as well
as storage size implications.
I suggest you try to take snapshot of magnolia 3.6 and test it ... we
are looking forward to your feedback.

Cheers,
Jan


> > From: "David Ziegler" <[email protected]>
> > Date: May 23, 2008 12:12:05 PM GMT+02:00
> > To: <[email protected]>
> > Subject: [magnolia-user] Huge Version Data
> > Reply-To: <[email protected]>
> >
> > I have a simple magnolia application and I am hooking it into some
> > external processes by periodically dumping the relevant content
> > as XML.
> >
> > The problem I am having is that the version data in the XML dump
> > is getting absolutely huge.   For example if I replace a list of 200
> > country names a few times, I can end up with 30 Mb worth of version
> > nodes in the XML dump.
> >
> > The version data is unintelligible, so I am unclear what it is  
> > recording,
> > but I have tried to be really careful about my update code (one  
> > transaction
> > etc).
> >
> > I am aware that I can ask Magnolia to dump without version data, but
> > all it is doing is using the stock jackrabbit dump (with version  
> > data) to
> > a file, then parsing it as XML and stripping the version data.    
> > Needless
> > to say this is very slow for a 30Mb file (and soon to be a 300Mb  
> > file if
> > things keep growing at this rate).
> >
> > I have removed the versioning command from
> > /modules/adminInterface/commands/website
> > but this did not stop the low level versioning.
> >
> > I am starting to play around with removing the "mix:versionable"  
> > type mixin at
> > node creation time, but am worried that magnolia might rely on the  
> > versioning
> > for "activating" nodes from the author to the public version.
> >
> > I am also considering using the jackrabbit APIs to strip out  
> > version data.
> >
> > One other possibility is to hack jackrabbit so that it has a dump  
> > option that
> > does not include version data.   (this is of course only workable  
> > if the version
> > data inside jackrabbit is not represented in a much more efficient  
> > way, and
> > therefore will not choke my system on it's own accord).
> >
> > So my questions are:
> >
> > - If i disable versioning, will this cause problems for Magnolia (esp
> > for activation)
> >
> > - Is there something I am missing that might be causing this  
> > explosion of
> > version data.
> >
> > - Is magnolia really fussy about it's jackrabbit version:  If I am
> > going the custom
> > jackrabbit approach, is it OK to use a newer version of jackrabbit
> > (starting a new
> > repository of course)?
> >
> > - Thanks
> >
> > David Ziegler
> >
> > ----------------------------------------------------------------
> > for list details see
> > http://documentation.magnolia.info/
> > ----------------------------------------------------------------
> 


----------------------------------------------------------------
for list details see
http://documentation.magnolia.info/
----------------------------------------------------------------

Reply via email to