I have to agree, the state of both SL and OpenSim is a royal mess. SL has already been forked, dozens of times. At SLCC '08 we spoke with Rob Lanphier about our announcement and presentation of the Meerkat viewer and were met with quite a bit of hostility. Linden simply did not want a competing viewer project at that time, and I believe that still holds true and is the reason Snowglobe was created.
The most popular 3rd party viewer used in SL, Emerald Viewer, is unfortunately run by people so proud and protective of their open source code that they refuse to have a public SVN, let alone allow any collaboration. The entire OpenSim community is so afraid of being sued by Linden that they won't allow their developers to look at the SL code for an entire 6 months before they can continue working on OpenSim. Even though the GPL explicitly allows a developer to study the entire SL code base and develop non-GPL code using knowledge gathered from the GPL project. The RealXtend developers fell right into that trap, and wasted a lot of time and money running concurrent development teams that couldn't look at each others code. And when the frustrations of their isolated development became too annoying, RealXtend started their own BSD viewer. Going forward, I think the best solution is to fork the viewer and OpenSim using distributed version control, while maintaining full compatibility with Lindens protocol, features, and servers. If we want the metaverse to progress and develop faster, we need to cater to the biggest metaverse communities out there, and develop clever ways to fill the voids that Linden refuses to. The Emerald Viewer team has shown was is possible when a large community adopts slightly ghetto hacked-in features like Client ID or chat encryption. We have the ability to store a lot of custom information on the SL grid in the form of textures (sculpts!) and notecards (client side AO!), so lets continue doing cool things with that ability. One thing that is certain is that the community is very willing to adopt these enhancements once they're made. I think it has become very clear that Linden don't want a community run viewer project, but a viewer project that they are in full control of that only works in their vision of SL. Patches for content import/export and grid hopping have been deleted from the Linden JIRA, further crippling our abilities to do cool things like being able to develop SL content entirely within 3ds max or offline in a local OpenSim server. The biggest hurdle will still be assembling a community of willing and able developers. Over the course of the two years that I ran the Meerkat Viewer project, we saw maybe 5 people step up and offer their time to the project, and of those only about 2 submitted something. It seems every developer has their own vision for the metaverse and is hellbent on running their own viewer project. We gave up with Meerkat and consolidated our work into another viewer project because we understood that developing the same things concurrently is a huge waste of time. I think that even now, the developers of other viewer projects waste half their time merging and patching in features from other clients! The end result is a handful of clients with more or less the exact same features. My hellbent vision of the metaverse is the Emerald Viewer team putting their code into Mercurial or Git and letting us have a go at it. The SL userbase has spoken, and Emerald is the most widely used community run project. Unfortunately the open source devs have had many years and opportunities to work together, and I don't see ours, or Lindens stubbornness magically changing any time soon. What is is going to take? On 3/14/2010 6:31 AM, Rob Nelson wrote: > I agree; In fact, people have been trying to communicate with Lindens > in some meaningful way (as is required in Open Source projects) since > the beginning of the open-sourcing of the viewer, but it seems that > Linden Lab seems more inclined to dictate what changes WILL be done > rather than gathering a consensus with OSS developers. See: hidden > SVN/HG servers, unreleased SL 2.0 beta source code, overall poor bug > turnover and triage in PJIRA (SVC-1509, anyone?), office hours vanishing > into thin air... > > Quite frankly, the idea of forking the entirety of SL (OpenSim and > viewers), as suggested by Morgaine, is looking quite attractive. At > least then the community can actually contribute without being shot > down, blocked by a wall of red tape, or chased off by rabid TPV > policies. We can then also contribute to the server-side of things > rather than waiting for the server goons to get around to adding or > fixing features. > > Fred Rookstown > > On Fri, 2010-03-12 at 21:26 -0500, Maggie Leber (sl: Maggie Darwin) > wrote: > >> ---------- Forwarded message ---------- >> From: Maggie Leber (sl: Maggie Darwin)<mag...@matrisync.com> >> Date: Fri, Mar 12, 2010 at 9:24 PM >> Subject: Re: [opensource-dev] Request for comments about >> llSetAgentEnvironment / SVC-5520 >> To: Soft Linden<s...@lindenlab.com> >> >> >> On Fri, Mar 12, 2010 at 2:31 PM, Soft Linden<s...@lindenlab.com> wrote: >> >> >>> A totally healthy open source project usually can be developed >>> completely in the open, and in a way that's aligned with everybody's >>> interests. But that takes an active commitment on all sides... >>> >> True enough. But I think there's widespread perception that the Second >> Life Viewer is already not a "totally healthy open-source project", >> and I don't think that perception can't be laid at the door of >> "obstructionism". >> > > _______________________________________________ > Policies and (un)subscribe information available here: > http://wiki.secondlife.com/wiki/OpenSource-Dev > Please read the policies before posting to keep unmoderated posting privileges > > _______________________________________________ Policies and (un)subscribe information available here: http://wiki.secondlife.com/wiki/OpenSource-Dev Please read the policies before posting to keep unmoderated posting privileges
