CiviCRM Community Forums (archive)

I suppose the real question is who's going to maintain them if they get cut loose from core?

Based on my previous discussion with Coleman, I suspect the three of a us generally agree that it's a good idea to move towards full modularization - and his message sounds like a first step in that direction. I'm not sure we've actually made the case for why that's a good idea. A few pros:

- To my mind, the big one is making the development process scale to include more developers at more organizations with more varied interests. This is done by giving folks more power to mix, match, fork, experiment, and re-assimilate.

- It encourages more design discipline. When it's obvious that one component is used by several others, one thinks harder about the software contracts.

- It loosens up the release scheduling - if some new feature in CiviGrant turns into quagmire, there is little risk it will hold up release of the platform. OTOH, if a user is frustrated by low priority bug that's fixed in trunk, then he can try the trunk version of the one relevant module - rather than updating everything.

- Everyone else is doing it - eg both Drupal and Salesforce have made the transition from applications to platforms.

Of course, there are downsides too:

- More attention goes into platform/infrastructure issues. Given fixed resources, this would mean fewer resources for deep thinking about use-cases that define our market. The question is whether modularization would subtract from fixed resources or help bring in more resources.

- Our tooling needs updates. For downloads/builds/dependencies, I like composer. It (can be made to) work with plain PHP libraries as well as modules in a framework. Since its PHP based, we could build a wrapper UI. Our phpunit setup will also need revision.

- Existing code would need to be reorganized. This can be done incrementally or all at once. The process will create issues or expose gaps in architecture, requiring us to refine interfaces.

- There's no guarantee that the benefits listed previously will materialize. The best-case is that consensus says they are likely to materialize.

(Note: Edited for typos. Originally typed this on a mobile device -- which was not easy.  )