How will upcoming changes affect older viewers?
I don't have a crystal ball, so this is of course speculation on my end, but there have been a lot of questions which i think i can attempt to answer with my knowledge of technology and the statements that Lindens made towards us.
I heard all V1 viewers will be banned!
OK, this is the worst rumor ever. We have been assured that this is simply not true. While it's totally up to Linden Lab to block access of their own 1.23 viewer to the grid because it incurs support on their end, there just isn't any plausible way they'd block us. See, feature-completeness is no way required for a third-party viewer, the only major requirement is that it doesn't harm the grid or other users, else text-only clients wouldn't be allowed, right? So please stop spreading this rumor, it's just wrong. You can expect viewers to suffer reduced functionality to a varying degree, and only Linden Lab 1.23 might get blocked eventually.
Will all V1 viewers become unstable and crashy?
No, Linden 1.23 will probably stay solid as bricks. All older viewers try to decode mesh data through the texture decoder, making TPVs among them prone to crash due to differences in texture decoder, but this is trivial to patch up, and you can expect this to be done in those V1 TPVs which are being updated. So if you keep yourself up-to-date, you might not get mesh, but you'll stay reasonably crash-free.
Will search stop working in V1 viewers?
There are two kinds of search used in V1 based viewers, classic search, and web search (the All and Showcase tabs of search floater, and perhaps the Groups tab). Currently used web search will be dropped soon and replaced, but as long as a viewer keeps updated, the developers can relatively easily make it use the newer one.
Then there's the classic search, which doesn't use an embedded web browser and shows convenient list of results on the left, and a single result on the right. All i have heard about this is "there are no plans to deprecate the underlying protocol", so it looks like it will stay in foreseeable future, but in fact it doesn't work perfectly any longer, some entries are just missing and can't be found even if you nail the query. We expect this to get worse with time, because Linden Lab has no genuine incentive to keep this protocol working well. For now, web search tabs can be used as an alternative, but if it comes to worst, it's no big problem to make search work exactly like in V2, it's just another embedded web browser window. The only thing holding us back is that not everyone would like it.
How will web profiles affect V1 based viewers?
Web profiles can be implemented as part of V1 based TPVs. However, we've taken a position there. A major, major V2 TPV, Firestorm, refuses to accept web profiles as the way to be! Bravo! And as long as they resist, and hold pressure on Linden Lab, we can resist too!
As Linden Lab has no genuine incentive to keep this feature working, there WILL be temporary breakage. In fact there is now, with friend permissions, SVC-7104. Feel free to make yourself felt on this issue by voting and subscribing to it.
If it comes to worst, it's no major effort to make profiles work like in V2, but that's not something we're going for.
How will Mesh content affect V1 TPVs?
Mesh is a fully new data type, which extends creator's freedom. It cannot be displayed with older viewers, or viewers based on older codebase. The possible crash issue is discussed above. In short, don't worry about that. You can expect Mesh to find its way on the grid veeeeery slowly, as upload costs for it are pretty huge.
My viewer, Singularity, has had the goal of integrating Mesh content - first display, then upload - for the last 7 months. We have been working hard towards that, and hope to have something to show soon.
Henri has started a rework of Cool VL viewer with a stated goal of supporting Mesh eventually.
Phoenix viewer has no immediate plans for Mesh content support, because essentially the whole team is working on Firestorm, yet they are interested to implement Mesh support in Phoenix too if they can do it without drawing many hands away from Firestorm.
While a clean work path showing one viewer developer implement or port a feature can always be leveraged by other developers to speed up their progress, Singularity's approach has involved reworking more than 120 000 lines of code beforehand to make Mesh integration easier, spread over half a year of work. Now, because of this, we don't feel like our implementation will fit easily within Phoenix, unless they repeat that whole work. But there is still a chance that an independent implementation makes its way into Phoenix, and if we can help them with knowledge we have gained, we certainly will.
How will Mesh affect older computers?
Owners of out of date computers have been a dorn in the eye of the hand of Linden Lab for quite a while due to their increased crash and disconnect rate. Official Mesh compatible viewers from Linden Lab do not support CPUs without SSE2 any longer, such as Athlon XP and old Sempron, and while borrowing any code from Mesh viewer, it will be very hard to retain compatibility with those. Note that Singularity has only had sporadic releases supporting these CPUs.
What other compatibility problems can be expected?
The unexpected ones! But looking at the kind of changes being made to the official Second Life viewer, hardly ones that can't be worked around.
One particular i'm mildly worried about is the new inventory delivery protocol. Snowglobe 1.5 based viewers include support for that (disabled for now), but it has its bugs which would need to be ironed out quickly if the old protocol suddenly malfunctions. Also, Viewer 1.23 would not survive that.