[sldev] RE: Speed of adopting open practices (Re: IJIRA and PJIRA)

[Matthew Dowd]

> My take on this is that ushering companies into the open source
> development model is a hard but extremely worthwhile problem to solve.


Don't get me wrong - I realise it is hard, particularly if you have an existing produce to maintain and develop (its like trying to change the engine of a car whilst still driving it etc.). I do feel, however, that in recent months we have seem some subtle but significant movement forward (the number of Lindens posting to this list, internal roadmaps being revealed prior to shipping code, possible due to cmake more frequent syncs from the internal svn to the public one, publication of some experimental internal code branches such as puppetteering and shadows, etc.).

Indeed Rob's post is itself a significant step: I think we had a similar discussion around a year ago, and the LL response concentrated on how far they had come etc. My frustration at the time was a concern that they were underestimating how difficult the process is, and how far they still had to go. An open admission that there is still a lot of work to do, is, IMHO, a major step forward! (and that doesn't belittle the progress already made).

The easiest (but by no means guaranteed, and still not easy) way to move is to just stop development for a month or three whilst you completely overhaul your entire development platforms, but few projects have that luxury. However, there was talk once of a SL version 2.0, but I think that was shelved in favour of just evolving the existing code? Let's face it, however, building transparent, fully maintainable code in large projects is hard; even if you follow the textbook there comes a point in most projects when the code has been re-architected/re-factored so many times that it needs a clean rewrite, and in reality most projects have some textbook "no-nos" in there somewhere - SL is certainly not an exception to that!

So a possible scenario would be for LL to start development of SL 2.0 from scratch as a completely open source project whilst maintaining the existing SL 1.0 code until the new version is ready for production. You already have part of the community for that work in the AWG community (and possibly even the OpenSim community - although there are issues re incompatible licenses there). I'm not saying that this is necessarily a model that LL *must* adopt, but one I think LL should seriously think about.

Matthew
_________________________________________________________________
The John Lewis Clearance - save up to 50% with FREE delivery
http://clk.atdmt.com/UKM/go/101719806/direct/01/