Paulo,

thanks. The pom.xml information is certainly useful as it proves how the 
tests were run. In most cases this will be enough. However, problems start 
happening when a later component (e.g. LARQ) has been tested with later 
snapshots than the current release. I think it is important that all 
components can run with a release for the other components (perhaps you 
already guarantee this?) Now, even if this is the case, it may still be 
tricky because approvals or adoptions by jena consumers (like myself) are 
done on a specific version release and updating it quickly because one 
component was not released before may be hard or even impossible.

So, from a consumer point of view, you ideally release everything together 
:-). It would also get rid of the various unaligned version numbers of 
different components, which are very confusing. Ideally, I just say "Jena 
2.7.1" (and includes IRI 2.7.1, ARQ 2.7.1, TDB 2.7.1, etc.) 

Simon




From:
Paolo Castagna <[email protected]>
To:
[email protected]
Date:
05/25/2012 12:59 PM
Subject:
Re: Next apache-jena release



Simon Helsen wrote:
> 4) LARQ: so far, we have not been using it, but we are thinking of 
> starting with that, so having it released with 2.7.1 would be nice. If 
> this does not happen, then, it would help if there is a clear indication 

> with what release a given component would be compatible (that request 
> would apply to any component which is not released with the "main" pack

Hi Simon,
knowing which module is 'compatible' with other modules is easily done by
inspecting the pom.xml of a module.

For example, if we look at LARQ 1.0.0-incubating's pom.xml file you can 
see that
it depends on:

 - Lucene 3.5.0
 - ARQ 2.9.0-incubating
 - TDB 0.9.0-incubating (for the tests only)

Same for other modules.

Answering questions such as would it work with any Lucene 3.x? Or, would 
it work
with any ARQ 2.9.x or any TDB 0.9.x? ... is another, non trivial, thing.
(Probably it is this you were referring to)

Some projects have conventions on increments of their minor version 
numbers (and
they promise that minor increments leads to drop in replacements fully 
backward
compatible). It also is not uncommon that bug fix releases [x].[y].[z + 1] 
are
just that and fully backward compatible.

I don't expect any 'big' and/or incompatible changes in LARQ 1.x.

My goal for LARQ is to keep its dependencies updated with changes in ARQ 
and TDB
as well as upgrade Lucene dependency as soon as a new Lucene 3.x release 
is out.
Longer term, I want to fix JENA-164 first, probably do nothing about 
JENA-242
and finally see if I find the time to actually work on JENA-17 for LARQ 
2.x

But, listening feedback (and complaints) from users is also what drives 
things.
For example, Lucene analyzers should probably be configurable in any 
serious
scenario, we recently have seen an example of this on the users mailing 
list.
However, the more we expose Lucene features, the more we get tied to it 
and
there might be issues in future when we want to allow using other systems 
such
as Solr, Elastic Search (although they are all based on Lucene ;-)).

Paolo

 [1]
http://repo1.maven.org/maven2/org/apache/jena/jena-larq/1.0.0-incubating/jena-larq-1.0.0-incubating.pom




Reply via email to