Moving some things from the Sandbox

[Marshall Schor <ms...@schor.com>]

I would like to move (at least conceptually, if not otherwise) some
things from the sandbox to be "mainline".  The first candidate I'd like
to do this with is uima-as.  Other candidates would be annotators that
seem stable, the simple server, and the CAS editor.

We could leave in the sandbox annotators that are not yet felt to be
stable, or are not "current" (such as the early version of a Groovy
annotator I added some time back, which has not kept up with changes
made to OpenCalais). 

This move, if we do it, would lead to the non-sandbox part of UIMA
having "multiple" kinds of things: Frameworks, Annotator components that
run in the frameworks, etc.  I think this is OK and is in line with what
the community wants.

I would hope we could find a very simple way to do this (in terms of
"work").  If it was a lot of work, I would hope we could "automate" much
of it so that in the future, things could be moved more easily.

What do others think?

-Marshall

Re: Moving some things from the Sandbox

[Jörn Kottmann <ko...@gmail.com>]

> I would like to move (at least conceptually, if not otherwise) some
> things from the sandbox to be "mainline".  The first candidate I'd  
> like
> to do this with is uima-as.  Other candidates would be annotators that
> seem stable, the simple server, and the CAS editor.

The Cas Editor is now split in the RCP and editing part. The editing
part could be distributed togehter with the other UIMA eclipse plugins.
To use the Cas Editor from eclipse one more plugin is needed to
provide settings and a type system. I did not worked on that in the  
past,
because I was very busy, but that changed and I could start with it.

 From my point of view, its more a question when we move the Cas Editor
Eclipse Plugin into UIMA ? Do we have already a date for the next  
release ?

Jörn

Re: Moving some things from the Sandbox

[Michael Baessler <mb...@michael-baessler.de>]

I like the idea to merge "stable" Sandbox components to the mainline UIMA. This is from my point of
view a step forward to a non-empty UIMA Framework enriched with built-in analytics. :-)
But I also think that if we do this, we have to carefully think about all side effects that this
move will have. We had a related discussion earlier on this list... please review
http://www.mail-archive.com/uima-dev@incubator.apache.org/msg04427.html

Currently the UIMA release concept is more like that we have a base UIMA framework (with some base
tooling) and we have addon packages with tools and annotators. If we merge now Sandbox components to
the base UIMA framework package we have to take care about the different "release cycles". E.g. an
annotator needs a fix - do we just provide hotfixes for that or a complete UIMA fixpack releases?

Do we want to track different versions internally for the annotator components of for the tooling?

I also think that if we merge Sandbox components with UIMA core we should take care of a separation
of the base UIMA framework jars, tools and UIMA components (annotators, consumers...). So that in
the UIMA package the users knows which jars are necessary for the stuff that he needs.

Another question is, how to organize the documentation so that the user is able to find the
paragraph he is interested in.

-- Michael

> I would like to move (at least conceptually, if not otherwise) some
> things from the sandbox to be "mainline".  The first candidate I'd like
> to do this with is uima-as.  Other candidates would be annotators that
> seem stable, the simple server, and the CAS editor.

I think we should additionally include the PEAR packaging projects for ANT and Maven.

> 
> We could leave in the sandbox annotators that are not yet felt to be
> stable, or are not "current" (such as the early version of a Groovy
> annotator I added some time back, which has not kept up with changes
> made to OpenCalais). 
> 
> This move, if we do it, would lead to the non-sandbox part of UIMA
> having "multiple" kinds of things: Frameworks, Annotator components that
> run in the frameworks, etc.  I think this is OK and is in line with what
> the community wants.
> 
> I would hope we could find a very simple way to do this (in terms of
> "work").  If it was a lot of work, I would hope we could "automate" much
> of it so that in the future, things could be moved more easily.
> 
> What do others think?
> 
> -Marshall

Re: Moving some things from the Sandbox

[Thilo Goetz <tw...@gmx.de>]

Marshall Schor wrote:
> I would like to move (at least conceptually, if not otherwise) some
> things from the sandbox to be "mainline".  The first candidate I'd like
> to do this with is uima-as.  Other candidates would be annotators that
> seem stable, the simple server, and the CAS editor.
> 
> We could leave in the sandbox annotators that are not yet felt to be
> stable, or are not "current" (such as the early version of a Groovy
> annotator I added some time back, which has not kept up with changes
> made to OpenCalais). 
> 
> This move, if we do it, would lead to the non-sandbox part of UIMA
> having "multiple" kinds of things: Frameworks, Annotator components that
> run in the frameworks, etc.  I think this is OK and is in line with what
> the community wants.
> 
> I would hope we could find a very simple way to do this (in terms of
> "work").  If it was a lot of work, I would hope we could "automate" much
> of it so that in the future, things could be moved more easily.
> 
> What do others think?
> 
> -Marshall

I think it would help if you could provide us with a concrete
proposal: what would our svn layout look like, what kind of
artifacts would we be releasing in the future, and what could
be a release chaining policy.  Once we have a concrete proposal
on the table, we can discuss that.

--Thilo