[OPEN-ILS-DEV] staring down the barrel of opensrf 1.0

[Dan Scott]

On 06/12/2007, Bill Erickson <erickson at esilibrary.com> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> OpenSRF 0.9 has been out for a while and much work has gone into OpenSRF
> trunk (message-level locale support, new JSON parsers and hinted objects
> format, to name a few).  I believe time for 1.0 is approaching...
>
> However, I'd like to propose we complete the following before it's released:
>
> * Add Makefile support for (optionally) installing the Java and Python libs
> * Add locale support to the Java and Python libs
> * Testing.. specifically with the locale code

On testing -- fully agreed, and it would be really, really nice to
have a basic set of unit tests implemented in one programming language
(pick a language, any language) that we could port to the other
reference languages. If we started by focusing on exercising the
locale support fairly exhaustively, we'll naturally end up covering a
lot of the rest of the framework as well. Ideally, we would eventually
have clients of one programming language interacting with the servers
in another language in a round-robin fashion, to shake out any
implementation bugs that might be lurking in dusty corners.

For 1.0, we could clean up the internal documentation for the OpenSRF
registered methods to ensure that they always have signatures with
descriptions and return info, along with accurate argc and param
descriptions (if applicable). OpenSRF/Application/Persist.pm just
lists api_name, method, and argc for all of its registered methods,
for example; OpenSRF/Application/Settings.pm is a little lean, too.

One more piece that would be appropriate for 1.0 would a migration to
autoconf / automake. But the current configure & install process
really isn't very onerous, so I wouldn't consider the lack of
autotools a 1.0 blocker.

On licensing: are we sticking with a GPL v2 license for 1.0, or are we
considering the possibilities of moving to GPL v3? 1.0 would be a
natural time to make the switch; there are still a small enough number
of contributors that rounding up sign-off could happen relatively
quickly and easily. I'm not particularly concerned about GPL v2 vs.
v3, in any case.

-- 
Dan Scott
Laurentian University