No Sansar

[Paul Hexem]

6 hours ago, Scylla Rhiadra said:

Who are you to say that "Barbie Dress Up Facebook" isn't an entirely appropriate way to use SL? Frankly, it's what drives a sizable chunk of the SL economy.

 

Because, like it or not, it's a 3D engine filled with content with bloated, million tri objects. LL needs to set people's expectations. "Second Life: We're laggier than Star Citizen. Better make sure your computer can handle it, because we can't be bothered to optimize things."

6 hours ago, Selene Gregoire said:

I'm able to run an GTX 1050 OC 2GB just fine. I do leave shadows off and reflections minimal most of the time since I'm usually building.

 

6 hours ago, Scylla Rhiadra said:

Yep, for sure, you're right! i think your case is not at all untypical.

And that's the point that Gadget and some others seem to be missing -- SL does work reasonably well for people without gaming computers. And it needs to continue to run reasonably well for people who don't have gaming computers, or you'll lose a sizable portion of the resident population, likely without attracting many new gamers to the platform at all in compensation.

The GTX 10xx series is exactly the kind of hardware animats and I are talking about. Even some of the 9xx series still hold up, and would benefit from advances in technology. 

[Pussycat Catnap]

15 minutes ago, Gadget Portal said:

Because, like it or not, it's a 3D engine filled with content with bloated, million tri objects. LL needs to set people's expectations. "Second Life: We're laggier than Star Citizen. Better make sure your computer can handle it, because we can't be bothered to optimize things."

Yeah.

This morning on my alt that's been resurrected to look at log cabins I bought a furry mesh body (Regalia). The entire body has less polygons than the left middle toenail on my preferred human body.

And it looks, animates, responds to physics and shape dials just as well...

SL 'human body' content creators are wacked out on crack... We got some badly optimized bloatware all up in this place as a result...

 

And yeah GTX 10xx boards are in the $100-200 range. Most of ya'll spend than on sex-Gatcha's weekly...

Edited April 2, 2020 by Pussycat Catnap

[Pussycat Catnap]

If we want SL to keep running well on boards like the 10xx-series... we need to slam the breaks on the polygon / vertex count count of what people can upload, and force a down-scaling of what's already in the wild... Bodies like Belleza, Maitreya, and Legacy - should all have LLs force cap their next updates to no more than 40,000 polygons - still 8,000 more than Regalia... and then give them 15-months before the current versions are pulled off the grid.

If we DO NOT do that, and we won't do it... then ya'll just need to get used to buying $400-500 GPUs like I did...

Edited April 2, 2020 by Pussycat Catnap

[Gage Wirefly]

15 minutes ago, Pussycat Catnap said:

If we want SL to keep running well on boards like the 10xx-series... we need to slam the breaks on the polygon / vertex count count of what people can upload, and force a down-scaling of what's already in the wild... Bodies like Belleza, Maitreya, and Legacy - should all have LLs force cap their next updates to no more than 40,000 polygons - still 8,000 more than Regalia... and then give them 15-months before the current versions are pulled off the grid.

If we DO NOT do that, and we won't do it... then ya'll just need to get used to buying $400-500 GPUs like I did...

You really don't need an expensive gpu for SL but it doesn't hurt. The power to lower the grind on your gpu is already there. I hard cap complexity at 60k and keep mine at or below that. It can be done and it works.If I encounter anyone over that limit, I don't see them and my humble gpu survives another day.  I still think LL should consider a hard cap on av complexity per user. It would go a long way in my opinion.

[Pussycat Catnap]

Maitreya:

Belleza:

 

And what furry folks get, Regalia:

Regalia is important because it's a furry body designed to let you wear a lot of human clothes - it's meant to match the shaping and rigging as much as possible for a lot of human fitmesh. It animates very well, responds very well to shape dials, has BETTER physics than any human body except Belleza... (for example unlike an unmentioned by me but popular human mesh body the ribcage doesn't move)...

In other words, you can get all that quality... WITHOUT any of that "lag" that causes people to need higher end GPUs...

And I suspect 34330 polygons is still HIGH compared to what some new high end video games use - but they also have limited animations.

 

Look also at the insane numbers for my human mesh head and my hair choices...

 

Edited April 2, 2020 by Pussycat Catnap

[RiffRegan]

Sansar has reinvented itself as a club/live music platform.

SL should maybe do the same. Just become clubs and shops selling clubbing gear.

[Nalates Urriah]

I am half-way through the thread and already want to respond to a bunch of posts..

20 hours ago, ChinRey said:

That would be the end of my days in Second Life. More importantly, I don't have any hard facts of course but I wouldn't be at all surprised if it would mean kicking out half the current SL population.

Raising the recommended system requirements doesn't lock people out. AFAIK, the Lab has always been highly cognizant of people with older machines. At the User Group meetings there is a lively contingent that want the Lab to stop supporting older computers. The Lab resists. 

As best we know they watch the user stats and base their changes on those. So, they know how many are on old machines and newer machines. To displace a large number of users would be a financial hit for the Lab. I can't see any reason for them to take that hit nor do I believe they will.

At some point software and hardware are simply so far out of date it becomes useless. Especially when it is at the expense of newer technology. I know a person that is still running on an old machine using 32-bit Windows-XP. Microsoft has stopped supporting XP, Vista, and 7. Many peripheral makers have stopped making drivers for Win-8. This iteration of SL's requirements are 8 & 10. But there are people running on Vista and 7. The individual on older equipment now has to figure out how to make the old stuff work. 

The Myst Online community has kept the servers running for a 2003 game that has had no serious update in tech for 17 years. There was a large enough contingent of people that knew the game tech and computers well enough that they could help people get the game running. As users lost interest knowledgeable supporters fell away and now one is on their own. My point is the company has to stay with the majority of users to remain profitable. The farther an individual is outside that 'main' demographic the more independent and capable they have to be. At some point people have to realize others can't be expected to take care of them. People are responsible for their life.

17 hours ago, Matty Luminos said:

Hell no. That would be me (and a lot of other people I know) kicked out of SL. They should be doing the opposite; ensuring that lower-powered devices like tablets and phones can access SL too, because nowadays those things are all that many people have to access the internet.

As above, changing the requirements doesn't lock people out. There are people running on Vista, 7, and 8.

They are building a client for Android and iPhone. Smartphones are way underpowered for 3D rendering. The latest generation is only getting close. But, this year they will have something. I expect next  year before it is anything I would WANT to use.

16 hours ago, ChinRey said:

[snip]

This is also a lot about recruitment. No remotely sane person will spend money on hardware just to check out Second Life. They will try log on with the regular everyday computer they have and if that doesn't work, they'll leave for good. It's only after they have to spent a bit of time there and decided they enjoy it before they might even consider buying special hardware for their new hobby.

That's a completely different thing, it's exactly the opposite in fact. Upgrading the software to run more efficiently should lower, not increase the hardware requirements.

Have you found such a person in SL? 😆

You right. Software optimization makes the software run better on all hardware. But, that isn't what users want. They want MORE. As things get optimized more stuff gets added in.

The Appliers era is coming to an end. They made understanding SL way too difficult. BoM is a multi-point upgrade on the server side intended to improve performance and simplify SL. As the uplift continues we will see more power on the server side of SL and the opening possibilities for taking load off the viewer, after years of taking the load off the servers.

12 hours ago, CaerolleClaudel said:

[snip]

The thing that finally got to me, though, is something that LL either can't do anything about or is incompetent at doing anything about, or some combination of the two. I have not really been able to find anything of enduring interest in SL for years. 

[snip]

Well, that isn't a Linden problem. SL is a world where the users create their own world. The Lab just provides the tools. You are knocking the wrong people.

12 hours ago, CaerolleClaudel said:

[snip]

Wrt to newbies, I am not convinced that a lot of new people are joining SL, or that the people who join stay. I rarely saw new people at the places I went. I do agree that a lot of the new people I see seem pretty lost. OTOH, I see a lot of month-old avis with full mesh bodies and awesome looks. I feel that SL does not make it easy to figure things out. Sure, they have the new resident orientation, but that is really basic. In spite of being here a long time when I got my mesh body, and even longer when I got my mesh head (and having experience with the mesh body), I had a really hard time understanding what to get and how to use it. I think LL has a Get Started page on the SL site, but it is Sign Up For Premium And Get A Linden Home and stuff like that. There is a wiki, which I find pretty awful. Maybe I have not done a good job of trying to find LL info, but anything I want to do means an internet search and a lot of digging to figure out what I want to do. I don't think I have ever run across guides and how-to things that just explain in simple terms how to do what I want. Same with finding things to do in SL. And the Destination Guide on the SL website is horrible, with mostly dead LMs.

[snip]

There is no convincing data I am aware of. What I can say is the signup numbers, counted by the SL computers, shows an average of about 10,000 people a day (300k+/- a month) signing up. March saw over 400k in signups. The daily peak concurrent user count average jumped from 51k to 57k in March. The number of daily unique logins is NOT published by the Lab. Ebbe says it something 600k per day. We and the Lab have wondered for years what is up with that 10k new signups per day. But obviously they didn't say. The Lab has an ongoing effort to solve the retention problem.

An OLD bit of trivia... the average SL users lasts 2 years.

As I responded to ChinRey, BoM is a performance and simplification change for SL.

When you find a dead link in the Guide do you report it?

11 hours ago, ChinRey said:

[snip]

There's a lot that can be done to increase everybody's fps without resorting to unrealistically high hardware requirements. I don't think I'm exaggerating if I said half the workload our cpus and gpus have to handle in SL never makes any difference to the actual appearance or experience and is a total waste of computing power. If anything, I'm underestimating the problem we have with content related bloated overhead. In addition there is also, as Animats said, a lot that could potentially be done to streamline the software to make SL more performant for everybody, regardless of how strong or weak their computers are.

But the first step in fixing the content overload problem is to document it though. LL has actually been working on that for over a year now as the first stage of Project ArcTan.  It is frustrating that it takes so long and it shows how far behind they fell during the six dark years of neglect. But it has to be done before any effective means to regulate the load can be implemented. LL has tried to put the horse before the cart (or maybe cure before the diagnosis is a better metaphor here?) several times. That has never worked well and it never will.

True. The problem is having time to do it. The Linden Lament is if it weren't for hackers they could do more.

The guy that works on ARCTan is also the guy that was working on Bento and BoM. In the middle of that we had a problem with 'disconnects' and a serious degrading of region crossing performance.

Where do you get the idea that SL has been neglected? They have been staffing up for most of the years you mention. At the UG meetings I am consistently meeting new Lindens.

11 hours ago, CaerolleClaudel said:

So, are walking in place and rubber banding related to low fps? I actually seem to experience both even when the fps meter shows 30-40 fps. I know that the lag at clubs is related to low fps, or assume it is.

Another frustration is landing at a club and seeing body parts flying through the air for several minutes. Everyone is naked for a while, too, but that isn't as disconcerting as having enormous heads zooming at me. And teeth, OMG...

Ah... no. Both of those are server-client interaction problems. So, either network problem or laggy server.

Yeah that flying body parts is a problem. Part of it from stacking numerous attachments on the same attachment point and the designer not bothering to correct the orientation when importing items.

There was a time when rendering naked was a big deal and lots of effort went into resoling that. It was better until we got mesh.

8 hours ago, CaerolleClaudel said:

[snip] ... I haven't found decreasing my graphic settings to help that much. [snip]

Another thing that is frustrating is trying to look at things while walking around; shopping is especially bad. IRL I can walk and look to the sides, but in SL your avi has to be facing things to see them (and don't get me started on stores having print scaled so small I have to zoom in to read it, and maybe even fiddle with my camera to get it close enough that I can zoom in enough). I guess I could have adjusted my camera so it looked sidewise, but that only takes care of one side. I realize this is a limitation of using a keyboard, but it is something that lowered my motivation to go explore or to shop. Probably VR lets you walk and look about at the same time, but that doesn't exist for SL, and I am not too interested in wearing a headset anyhow.

Tweaking the viewer for performance used get a lot more discussion. It is an arcane art. Settings for best performance and personal taste are different for each computer/person combination. There are tutorials on 'how to' get the best... 

Not true. I walk parallel to the wall while looking at it... I use mouse-steering, left-click-hold on the avatar and press A or D. I use that in combination with W and S and the mouse. If it is crowded I switch to Flycam mode and fly the camera to my right ot left while looking at the goodies. No camera adjust needed. A SpaceNavigator makes it REALLY NICE.

Headsets suck. Messes my hair and runs my makeup.

Edited April 3, 2020 by Nalates Urriah

[Ardy Lay]

1 hour ago, Nalates Urriah said:

There was a time when rendering naked was a big deal and lots of effort went into resoling that.

Now this is a feature.  😉

[Mollymews]

21 hours ago, ChinRey said:

That would be the end of my days in Second Life. More importantly, I don't have any hard facts of course but I wouldn't be at all surprised if it would mean kicking out half the current SL population.

that would be me as well.  My computer is about 9 years old now. I recently upgraded to a 1050Ti from a 660GTX but still only have 16GB RAM

my computer trundles along (relatively) and I am ok with this.  I was actually ok with my 660GTX and then 1080x1024 screen. Then I saw this really good 2K curved screen that this other person had. And I was like wooo! sharp amazing vibrant colors and lots of extra screen space. So I bought one and decided to get a 1050Ti to drive it.  I never got any higher like a 1060 because I would have had to get a bigger power supply and a new motherboard as well. So I never

i didn't buy my new screen to use on SL. I bought it for the RL work I do on it and get paid for. Being able to use it for SL is just a bonus

the problem with putting arbitrary hardware limits like 1060 minimum, is that this can become a perennial. This year 1060. next year 2060, year after 3060, year after that 4060 and so on. For me to justify the cost of this kind of ever increasing upgrade cycle then I need a RL reason to do it, not just a SL reason

[Lyssa Greymoon]

2 hours ago, Mollymews said:

the problem with putting arbitrary hardware limits like 1060 minimum, is that this can become a perennial. This year 1060. next year 2060, year after 3060, year after that 4060 and so on. For me to justify the cost of this kind of ever increasing upgrade cycle then I need a RL reason to do it, not just a SL reason

Don’t worry, at the rate Linen Lab updates their recommended system specs, if they put the GTX 1060 on it this year, it will still be there when the GTX 8060 is released.

[Maitimo]

6 hours ago, Nalates Urriah said:

At some point software and hardware are simply so far out of date it becomes useless. Especially when it is at the expense of newer technology. I know a person that is still running on an old machine using 32-bit Windows-XP. Microsoft has stopped supporting XP, Vista, and 7. Many peripheral makers have stopped making drivers for Win-8. This iteration of SL's requirements are 8 & 10. But there are people running on Vista and 7. The individual on older equipment now has to figure out how to make the old stuff work. 

My system isn't even that old, it's just "budget-economy level", because I have a low-income job and other things, like rent, have to take priority over upgrading it. I'm running Win-10 but my graphics card is only a 960GTX and I have only 8GB of RAM. (But it does have an 8 core processor 😄).  And most of the time it performs more than adequately in SL; I generally run Firestorm at Medium to Med-High settings and I typically get a fps of 30 or better which is not a problem.  I can even run Black Dragon without any problems. It's only at busy clubs, with 20+ high-complexity avatars, where I run into difficulty. Then my fps drops to about 5, even on the very lowest settings. The only way I can survive them is to park up in a safe place and de-render everyone.

I actually already have a laptop with better specs than my desktop, but I hate using it for SL on that tiny screen.

[ChinRey]

5 hours ago, Nalates Urriah said:

Where do you get the idea that SL has been neglected? They have been staffing up for most of the years you mention. At the UG meetings I am consistently meeting new Lindens.

"The six dark years" refers to the period from 2008 to 2013, after the original LL crew gave up and before the current regime took over. The current LL is anything but flawless but nobody's perfect and considering where they had to start, they've done a remarkably good job. If there had been a world championship in code cleaning, Oz and his crew would have been my favorites.

More workers doesn't neccessarily mean better work. Generally a small well coordinated team where all pull together will achieve much more than a large group of people with no direction or clear goal.

[Mollymews]

27 minutes ago, ChinRey said:

 If there had been a world championship in code cleaning, Oz and his crew would have been my favorites.

mine too

i have never met Oz Linden or any of the people on the team since Oz Linden came to the Lab.  I just know their product and it is good compared to what the product was like prior

this is not to say that the prior team/teams weren't good or never did their best in the circumstances confronting them (different times, different obstacles and challenges), they did do the best they could and can't be faulted for that

but still, the product speaks for itself and from a consumer/customer pov this is pretty much all that matters, at least in terms of opening wallets and purses

 

 

Edited April 3, 2020 by Mollymews
ses

[Drayke Newall]

I've mentioned many things I would like to see done to or in SL so wont repeat them here, however one thing I think they should do I haven't mentioned is due to the sim visitor limit, they should make it so that you can sync sims. This would mean large events could be held such as a live music festival or speech event etc whereby multiple visitors to the region can see the same event live without restriction on numbers.

If that isn't clear basically have the stage and performers on a sim and then have it so that that main sim is linked to multiple sims that are not visible to the other sims allowing 1000's of people see the same event live and not just 70+. Hard to explain but hopefully I explained it so people understand.

They did something like this last year in the game Fortnite  https://www.theguardian.com/games/2019/feb/03/marshmello-fortnite-in-game-concert-edm-producer

Edited April 3, 2020 by Drayke Newall

[ChinRey]

1 minute ago, Drayke Newall said:

I've mentioned many things I would like to see done to or in SL so wont repeat them here, however one thing I think they should do I haven't mentioned is due to the sim visitor limit, they should make it so that you can sync sims. This would mean large events could be held such as a live music festival or speech event etc whereby multiple visitors to the region can see the same event live without restriction on numbers.

Simulator instancing was one of Sansar's big selling points at first and it's probably the main reason why they managed to find a buyer for it. They must have considered implementing it in Second Life too but apparently there are technical reasons why it can't be done.

They do have another solution to increase capacity for larger events though, sim corners:

Four times the capacity of a single sim and I've yet to hear of an actual event that needed more than that. But of course, nobody can justify paying tier to keep four regions on standby for the rare occasions such a solution is needed so it's only LL themselves who can afford to have them.

[Drayke Newall]

5 minutes ago, ChinRey said:

Simulator instancing was one of Sansar's big selling points at first and it's probably the main reason why they managed to find a buyer for it. They must have considered implementing it in Second Life too but apparently there are technical reasons why it can't be done.

They do have another solution to increase capacity for larger events though, sim corners:

Four times the capacity of a single sim and I've yet to hear of an actual event that needed more than that. But of course, nobody can justify paying tier to keep four regions on standby for the rare occasions such a solution is needed so it's only LL themselves who can afford to have them.

True, however that does have its limitations in that you cant have front view syncing and also limits drastically the area in which people can stand to see the stage. Also with most music events 2 sims will be looking at the artists or performers backs.

[ChinRey]

47 minutes ago, Mollymews said:

this is not to say that the prior team/teams weren't good or never did their best in the circumstances confronting them (different times, different obstacles and challenges), they did do the best they could and can't be faulted for that

I don't know for sure but yes, it's my impression too that the developers shouldn't be blamed for the misery. Lack of communitcation and coordination, unclear directions, poor situatin awareness - the responsibility for all of that lies with the top management, not middle management or the workers on the floor.

[CaerolleClaudel]

4 hours ago, ChinRey said:

"The six dark years" refers to the period from 2008 to 2013, after the original LL crew gave up and before the current regime took over.

Interesting. Most of that period is when I loved SL the most. I guess at that time I was more focused on the social aspects than on technical aspects, though I really really really hated lag. But that was pretty much from day one in 2007 through the last time I was in SL.

[Nalates Urriah]

12 hours ago, ChinRey said:

"The six dark years" refers to the period from 2008 to 2013, after the original LL crew gave up...

I came in during 2008. By 2009 I was participating in the various UG groups facilitated by the Lindens. During that time we were getting lots of new features. I see no reason to think of 2008-13 as 'dark years'.

[animats]

Just watched Lab Gab, on the "cloud uplift". No new viewer is planned. Oh well. It was mentioned that inventory management has already been moved to "the cloud". That's progress. It's probably the most complicated non-world thing the simulator does.

Originally the region sims did the whole job of running SL. Then asset serving (meshes, textures) was moved out, which is good. Now inventory. Over time, more stuff that doesn't involve the live 3D world may move out - chat, money, Marketplace interface, groups, LSL compiling, mesh uploads, etc. Those are more like standard web server tasks, and will be easier to maintain if they're not entangled with the main sim code. Also, AWS's scaling features are good for such things. Looks like they're going that route.

[ChinRey]

1 hour ago, animats said:

Originally the region sims did the whole job of running SL. Then asset serving (meshes, textures) was moved out, which is good. Now inventory. Over time, more stuff that doesn't involve the live 3D world may move out - chat, money, Marketplace interface, groups, LSL compiling, mesh uploads, etc. Those are more like standard web server tasks, and will be easier to maintain if they're not entangled with the main sim code. Also, AWS's scaling features are good for such things. Looks like they're going that route.

That sounds like really good news and yes, those are functions that should be very well suited for cloud hosting.

[animats]

1 hour ago, ChinRey said:

That sounds like really good news and yes, those are functions that should be very well suited for cloud hosting.

Yes. Also, Oz Linden has said that the mobile viewer does not put in you in world; it just does chat, groups, etc. Today's Lab Gab  said that the "cloud uplift" is being done one piece at a time, in small pieces. That suggests that the sim server code is being split into "user stuff" and "3D world stuff." and put on different servers. The 3D viewer  needs to talk to the region sims, but the text-only mobile viewer does not. When the mobile viewer comes out, it's likely that anything it can do has been moved out of the region sim processes.

This is a good thing. The less the region sim servers need to know and do, the faster they can run. Less excess baggage dragged around as an avatar moves from one sim to another. The region sim servers don't really need to know what you're wearing to do their job. They  just track that so they can tell the viewers. To the sim, you're a cylinder with rounded end caps. If the region sim servers can pass the buck to the user server that's handling your current login when some viewer wants to know what you're wearing, teleports and region crossings become much lighter-weight operations. Being overdressed stops impacting the region sim servers. Don't know if they're going that far.

[ChinRey]

21 hours ago, Nalates Urriah said:

I see no reason to think of 2008-13 as 'dark years'.

I do.

I joined SL in 2013, right at the end of that period and I too think of that as the best days of my Second Life. That's partly because it was all new and exciting of course but also because everything seemed to work better back then.

But even in a virtual world there is a difference between seeming and being and a shiny surface doesn't mean all is well underneath.

Think of it as a pimped up car. The owner spends a lot of dough on a fresh new custom paint job, a stereo that plays loud enough it shows up on seismograph all over the world, more extra lights than Genesis' 1977 stage rig (they used the landing lights for a jumbo jet for that), hubcaps so bright you go blind if you look at them straight, the whole package. But he never opens the hood. He never changes oil, never checks the coolant fluid, never replaces the worn out break shoes and rusty suspension springs or the clutch he burned out long ago.

It's always tempting to skimp on maintenance, especially on parts that don't show because... well because they don't show. Not until later that is. Every dollar you "save" by neglecting maintenance will eventually cost at least ten dollars in repairs and that's the price we are paying now.

---

I could probably write a long list of complaints here but I think I'd rather mostly focus what Linden Lab and various Lindens actually have said. They will never and should never openly criticize LL the way we as customers are allowed to do of course but sometimes they slip and sometimes they can't avoid it.

---

When opensim appeared, there was a lot of talk about releasing SL's server software as open source. One Linden (I can't remember his name and I can't find the source right now) explained that this was impossible because the code was so full of monoliths. For those not familiar with anti-patterns, monoliths (aka stovepipes) are multiple separate pieces of code doing the same job. A good example is the way normal maps are handled. There are three totally different parts of the software that do that, one for the legacy bumpiness maps, one for the system water and one for the "new materials" maps. Why? Because LL kept piling new shinies on top of than intergate them into the existing software.

---

The opposite anti-patterns to monoliths are spaghetti code and action by distance. That's when you have unwanted dependencies, some code is doing totally different things in totally different parts of the system. Imagine if the light switch in your bedroom also switched your water heater on and off and you get the idea. This is of course one of a programmer's worst nightmares. The two most common causes of spaghetti code are copy-and-paste programming and "rockstar" programmers but there are others too.

During the early stages of the Sansar, Ebbe's explanation why they started a brand new project rather than upgrade SL was simply "too much spaghetti in the code".

One rather comical example from SL. In 2013 they launched SSB (Server Side Baking). For those who have joined later before that the textures for each layer of system clothing, skin and tattoos was downloaded separately and merged by the viewer. That caused a lot of unneccessary overhead throughout the system of course and baking it all server side was a very good (and long overdue by then) idea in itself. I lived on one of the RC sims they first launched it on at that time and suddenly I saw the skin texture for my avatar plastered all over the sky. Somehow they had managed to mix up the code handlings textures for avatars and windlight. How that is even possible? I have no idea.

---

Copy-and-paste programming is an anti-pattern in itself. I suppose there's no need to explain what it means. It's not always a bad thing, you can save a lot of time by reusing old code rather than write new one from scratch but it can and will backfire badly if you're not careful. In addition to the spaghetti, typical symptoms of poor copy-and-paste coding are recurring bugs and bloated overhead. LL used copy-and-paste coding extensively throughout the six dark years and I don't know where to start listing all the problems that has caused.

---

Coordination was – and still is – a problem with LL.

I once had a long chat with Michael Linden about the Wilderness – that is, he talked an I listened (yes, I can do that too if I want to!). He was very proud of that build and for good reason, it's one of the best early mesh builds I've seen in SL. But he also told how he and the Moles had to learn mesh from scratch. He was proud of that too, again with good reason – I just wish they hadn't stopped there. But although he didn't think of it, this also raises a big red flag.

Back in the earlier days of SL, Cory Ondrejka and Eric Call, heads ot the developer Lindens and content Lindens respectively, worked clsoely together, coordinating their works and leaning on each others' formidable expertise.

That ended late 2007 when Cory was kicked out with no previous warning. When LL started working on mesh, their content creators seem to have been kept out of the loop entirely. That meant the developers had assumptions and to some degree confusing and conflicting demands from users - all with their own special interests – to determine what was important and what wasn't. Then, when the Moles started to build with this new material, they – just like all other content creators - had to try to reverse engineer the whole thing with little or no help from vague and incomplete documentation written in the divine language of programmerish that mortal men and women were never meant to understand.

---

I was trying to explain why I call 2008-2013 the six dark years of SL. It's already one and a half page and that ridiculously long for a forum post. I hope it's enough to explain but if not, let me know. I've barely started and have lots more.

Edited April 4, 2020 by ChinRey

[Nalates Urriah]

14 hours ago, ChinRey said:

I do.

I joined SL in 2013, right at the end of that period and I too think of that as the best days of my Second Life. That's partly because it was all new and exciting of course but also because everything seemed to work better back then.

So... SL was a dark age before you got here? And you know that because someone gave you their opinion of that time or what?

The JIRA has a Reports section. In the 2009-13 time period I was into the then called 'Office Hours' the Lindens had. We call then User Groups now. I used those JIRA reports to see how many feature requests and bug reports were bing filed and how quickly - filed vs resolved. THe Lab was keeping up with fixes and new features were getting added at a good pace. SL users were claiming nothing ever got fixed. Just as they do today. I looked for the truth.

In late 2009 new Windlight settings, mesh upload, deferred rendering and other features were being added, the relationship with content creators was being modified per requests, XStreet (the marketplace then) was being improved, script limits improved... I pulled these changes from just a couple of months in late 2009. Then there is Changes Coming to Second Life – 2010 outlining what we were expecting for 2010. 

In 2012 I wrote my first update of what had happened in SL during 2012. (Ref) Then again in 2013 (Ref). 2012 and 2013 were not exceptional years of SL history. Development and progress were on going from 2008 when I came in to now. Not everyone liked the ideas coming out of the Lab then but their opinions don't make the 2008-13 period a 'dark' age.

You can compare the 2012 & 13 articles with later articles like the 2015 June Review. I did other months and years. The pace of development has been pretty consistent from 2008 to now. 

What has changed is how the Lindens talk about what they are doing and plan to do. They have moved from open to secretive to open to somewhere in between and back and forth... it changes. Various Lindens are better with their UG meetings than others.

There are periods when the Lindens take on huge projects, like Uplift, and feature addition slows. Some features they underestimate, like EPP now months past the expected completion date. But, the work pace is reasonably constant.

Trying to criticize the Lab's management is easy but in reality it is hard to find other 'games' like SL that have lasted as long as SL or any that produce income for users on the scale that SL does.

 

[animats]

22 hours ago, ChinRey said:

The opposite anti-patterns to monoliths are spaghetti code and action by distance. That's when you have unwanted dependencies, some code is doing totally different things in totally different parts of the system. Imagine if the light switch in your bedroom also switched your water heater on and off and you get the idea. This is of course one of a programmer's worst nightmares.

Quite aware of this, having been coding some improvements to the viewer lately. It's not unmaintainable, but due to technical debt, maintenance is harder than it needs to be.

I see some real promise in the way the "cloud uplift" is going. The uplift clearly moves more functionality out of the sim code to other servers. I'm guessing here, but the uplifted architecture might go like this:

You log in, and some load balancer assigns you a slot on a "user server". There are many user servers on Amazon Web Services, each serving maybe a few hundred users. More user servers are started and stopped as more or less users are logged in.  That's what AWS is good at, dealing with variable load. The user server is running part of the code of today's sim servers. but not the part that does the 3D world. Your inventory, preferences, chat logs, groups, friend lists, Marketplace stuff for sellers, etc. are loaded from databases.

The user server then finds out where you were in the 3D world. and sets up the connections to the proper sim server for that region. The sim server is running part of the code of todays's sim servers, but minus the stuff the user server does. That includes all the stuff listed above, which the 3D world does not need to know about.

Anything which can be done in the user servers takes load off the sim servers. How much can be done in the user servers? Everything listed above, at least. It's noteworthy that Oz Linden has said that inventory has been moved out of the sim servers, because that does have some in-world connections.

A big win would be to move appearance, clothing, wearables, and attachments out of the sim servers. The sim server has no interest in what you're wearing. Attachments don't collide. That info is between the user server and the viewer, with help from the asset servers where the mesh and texture files live. Going further, all worn scripts could run in the user servers. Sometimes they'd have to ask a sim server for some info or ask it to move something. But most worn scripts are for clothing adjustment, and they need zero interaction with the sim server. LL may or may not be doing that. But it's a big win if done.

This leaves the sim server doing mostly physics, object movement, and object scripts. Most of the load associated with avatars is elsewhere. Avatars would also have far less excess baggage to be copied from sim to sim as they move. So the impact of teleports and region crossings would be much less. The sim servers rarely have much trouble with object movement and physics load. Most of the load is scripts, as anyone who looks at the performance window in the viewer is aware.

The payoff is more avatars per sim before things slow down. Maybe many more. A thousand avatars in seats is just a thousand stationary prims to the sim server. A huge problem with LL as a social system is that too many people in one place makes it choke. Now this basic problem can potentially be overcome. SL can have big concerts. Even Fortnite can't put a few thousand people in one place. They have to fake it with multiple copies.

I don't know if this where "uplift" is going. This is me speculating on how LL is implementing what they've said they are doing. But if there's a split between sim servers and user servers, which it looks like there is, this is where it leads. It's a way to make SL scale better.

(Yes, viewer overload. See my other posts on that.)