Re: [Exist-development] registering modules at runtime

[Dannes W. <da...@ex...>]

Hi,

On Mon, May 31, 2010 at 2:00 AM, Joe Wicentowski <jo...@gm...> wrote:
> Evgeny - Cool.  I am looking forward to giving these new functions a
> try.  I think Dannes asked for unit tests too - something I think is
> necessary to ensure the functions can stay in trunk.  Could you
> contribute unit tests when you have the chance?  Also is there a
> deregister/unmap function, or only a register function?

jUnit tests are really required. Probably if no tests are provided
we'll need to remove the new functions again.

This thing is: I am pretty sure the 'good weather behavior'  is okay,
but since this function introduces some dynamics, which can be rather
complex, there are risks.  Scenarios like

- a module is registered but does not exist. What should happen? How
is the user informed when things do not work?
- what happens if a module is registered successfully, is successfully
used for some time and then it is removed or unmapped? How can we deal
with this in a production environment?

All in all I think we are not done yet, there is work to do. Maybe the
function is introduces too quickly. I did not see a 'go' from the core
dev team, only 'some' discussions for one or 2 days on the ML.. Before
adding new functions we really need to have a 'go' of e.g. wolfgang.
We really need to keep control on new functionality....

cheers

Dannes

-- 
eXist-db Native XML Database - http://exist-db.org
Join us on linked-in: http://www.linkedin.com/groups?gid=35624