The Viewer Situation on OpenSim

[Snoots Dwagon]

Opening up a can of worms. Please pardon the somewhat frank, no-punches-pulled post here. This is a "for what it's worth" opinion post. : )

This details history with the Firestorm project. I appreciate the Firestorm viewer. It's my favorite. But (at least) two prominent bugs have been left unfixed for years, and undoubtedly far more than that. These two major issues are: 1) Slow on-demand* texture rezzing and 2) Slow object Content retrieval that can take 5, 10, 20 minutes or more (other viewers do so in seconds, so this is a Firestorm-specific issue).

* Note: "On-demand" texture loading refers to trying to see a texture that was just uploaded to inventory, or applying a texture from inventory to the face of a prim, or shoppers trying to get a vendor to rez a product photo. Direct user-to-system interaction.

Builders work with dozens of textures at a time. Shoppers view hundreds of textures in vendors. It is understandable that 15-24 second texture loads can extend the time involved in a building project extensively and delay a shopping experience to the point of buyer frustration.

So, discussing this earnestly: The effect of Firestorm on the OpenSim project and the dedication / intent of the Firestorm team in supporting OpenSim specifically as a well-established platform.

When we experience a major problem, is the cause Opensim or Firestorm or both? How do we find out if the Viewer dev team is less than responsive to inquiries?

Is Firestorm dedicated to the OpenSim project (last I heard they had one OpenSim dev) or is the team tunnel-focused on the highly-profitable Second Life platform and Opensim considered a second-class system?

That perhaps is a question that deserves attention by the entire Opensim community... as there are more Opensim regions than there are Second Life regions. OpenSim users number in the tens of thousands and may even exceed Second Life (it's difficult to tell from day to day, especially with Alts on both platforms, as well as the hypergrid).

There is no doubt Opensim is huge. The platform needs a more attentive viewer development team, either through Firestorm... or that failing, other options. If we're going to have all these grids with tens of thousands of regions... we need to be very serious about a Viewer system that is responsive specifically to Opensim needs. The Viewer is our access to all of these worlds.

THE REASON THESE THINGS AREN'T GETTING FIXED AT VIEWER LEVEL
Two years ago I reported on a Firestorm ticket an issue dealing with Object Contents loading with extreme slowness (a serious bottleneck). The link is here: https://jira.firestormviewer.org/browse/FIRE-21697

This bug was first reported to Firestorm 2017/Sept 25. If one reads the comments following, it is easy to see the runaround response Firestorm gave me on this issue (finally telling me to take the problem to Linden Lab, although I don't usually use that platform). At that point I stopped dealing with that nonsense. Volunteer or no, there is a certain promissory responsibility and expectation-- at the very least to recognize and log serious bugs when they exist and take responsibility for fixing them.

The above Ticket refers to another ticket which was filed a year before, regarding a similar subject. Results: The Contents lag problem was brought to Firestorm's attention some four years ago-- and is still not resolved. Firestorm's response/excuse: That's Linden Lab's problem. They need to fix it.

I have to ask: Is Opensim a Linden Lab company? Is Firestorm a Linden Lab company? Are they really dedicated to Opensim as a separate entity, or is this platform of secondary importance in their minds? Does Firestorm realize how large and extensive Opensim is?

It would seem that grid owners are going to have to use whatever pull they have with Firestorm to get these types of problems fixed, or create their own workaround to a specific bug (if that's possible). Kitely has been known for the later: updating their system to work around a severe Viewer bug.

Object Contents and on-demand Texture Loading are two data-verified severe bottlenecks. Why is Firestorm telling Opensim that is Linden Lab's problem? Surely they are aware that Opensim is not Linden Lab.

Firestorm has a known history in dealing with such things. Even when detailed specifics and undeniable proof are provided, the dev team has often denied the problem or shoved responsibility off on others.

I mean no insult to the Firestorm project. They're understaffed, overworked, and regularly ridiculed for bugs. I get it; that can be very frustrating. But they've chosen to work without funding (people have offered to set up a donation system so Firestorm could have paid devs). One reason I left Second Life was because of unresponsive support from Linden Lab. I didn't come to Opensim to find more of the same. Support is one reason I support Kitely. If Firestorm doesn't have that same commitment, perhaps Opensim needs to consider professional development of its own viewer.

A POSSIBLE SOLUTION
Perhaps one solution to this would be for the major grid owners (as well as businesses that profit from this platform) to get together and set up a professional, paid (non-volunteer) Viewer Development Organization to fund a hired coder team to produce a reliable and stable Opensim-specific Viewer. Paid coders would be under direction to fix that which needs to be fixed. Weary and under-appreciated volunteers would not be part of that scenario. "I don't want to work on that" would not be part of that scenario. Coders would be assigned debugging and development tasks and being paid, expected to perform them. This would be a significant improvement in solving years-long mega-bugs... and finally cut the apron-strings to the Second Life viewer and Linden Lab.

Opensim is growing, it has already surpassed Second Life in scope. Perhaps it needs its own Viewer dedicated to the specific needs of Opensim worlds.

[Ada Radius]

It would also take professional fundraising and a nonprofit entity. Blender Foundation is a good model for that - they started as a for-profit, didn't make it, regrouped as not-for-profit, and now have a small core of professional devs, hundreds of volunteers contributing code on things they care about. corporate and private sponsorship.

[Selby Evans]

You may want to catch and discuss this presentation at OSCC
06 DEC SCENEGATE, ECHOVOICE, AND IMABOX
7:30 am - 8:30 am Keynote Auditorium
https://conference.opensimulator.org/schedule/