cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joerg Heinicke <>
Subject Re: Versioning XML Schemas
Date Sun, 01 Jul 2007 14:37:15 GMT
On 01.07.2007 15:46, Grzegorz Kossakowski wrote:

> I would like to add some more documentation to 
> cocoon-configurator-1.0.1.xsd[1] but I'm not sure what is our policy for 
> versioning XML schemas. AFAIK, 1.0.1 version has not been released yet 
> so I can modify it freely, right? When it will be officially released? 
> When cocoon-spring-configurator is released?

Yes, I think it should be released with the block it contains it. I 
don't know if its versioning needs to follow its containing block 
though. Not every release of the block makes changes in the schema 
necessary. Despite steps in the versioning it will probably be easier 
though to follow the block's versioning.

> What is more confusing we already have schema published at [2] without 
> any mark that this is a snapshot version that is likely to be changed.

IMO that's no problem. See below ...

> Do we have any policy on versioning xml schemas? Shouldn't we have a 
> cocoon-configurator-1.0.1-SNAPSHOT.xsd published and rename it just 
> before we release cocoon-spring-cofigurator? This way we would always 
> know when the version is locked down.

I'm against the SNAPSHOT suffix. A DTD or schema should never be 
retrieved from remote, but always from a local version (via xml 
catalogue or whatever). For somebody doing the latter it should be 
obvious whether it's released or not since he works with the released or 
the snapshot block. IMO it's just too much work to change all the 
references from the snapshot version to the release version if they 
differ in name.

Spring handles it the same way [1]. They did not even increase the 
version number on changes [2]. Important is probably the backwards 
compatibility. In that particular case: An XML written against Spring 
2.0.2's AOP schema works in 2.0.3 as well despite the additional 
attribute. No idea yet when it is better to force the user to update his 



View raw message