Musings on 1.8

I was pretty amazed to see 1.8 is to be released this coming Tuesday (13th).

I've only checked out the preview a couple of times so far, assuming, incorrectly, that it was going to be comprehensively tested prior to release...
So I felt no particular urgency. Especially seeing as it was (and still is at the time of writing) impossible to buy land in 1.8, and thus impossible to play with the P2P tools that are of great concern every landowner.

'Landowner': Someone who pays Linden Lab land use fees in order to 'own' land in SL

So, what else is there in 1.8?

Fixes for the speed we all lost in 1.7? Who can tell? It's running, just like 1.7 Preview, on a tiny and virtually uninhabited grid. So it seems fast. But then so did 1.7 Preview...

Fixes for the texture loading/cache issues we are all suffering from? Doesn't look like it. Moving back and forth between locations I'm still waiting for textures - which SHOULD be cached on my HD after the first visit - to load...
To tell the truth I don't really know what's going on with textures, today in 1.8 I saw a number of textures displaying the default grey and white with the word 'LOADING' superimposed in white. Doesn't sound much like the texture loading issues have been resolved does it?

Map improvements? Well it's still sluggish, it's still got that useless 'got to be able to spell the sim name' Region Locator, and there are Land for Sale beacons all over it when there's no land for sale. Great.

So what great advances DO we have in 1.8?

EYE CANDY! Ripply water. OMG! Just what we needed to take our minds off the performance issues that have been evident since 1.7...


I've often wondered about LL and it's management. What I'm seeing isn't increasing my confidence one little bit.

I wonder if they realise that 1.8 is probably their last chance as far as many customers are concerned?
It needs to deliver fixes - not gimmicks
Even more importantly it needs to be reliable and work straight out of the box. Another fiasco of 1.7 levels simply will not be tolerated by many.

So WHY is it being rushed out with very little testing and no obvious improvements to 1.7?

Bolting P2P onto 1.7 and chucking in Ripply Water for the 'Ooooh - Shiny!' brigade DOES NOT constitute a major update in my opinion, and doing so without proper testing and, more importantly, without resolving the remaining 1.7 issues is asking for trouble.

Carry on.

I'm not a "sky is falling" kind of girl, but SL has reached a critical point in customer confidence IMO. Give us performance fixes.

Poor management + Lack of Funding + Poor Marketing = Where We Are Today.

The only real question is 'Who will own this next and will they actually make it better?'

I fear the answer to that question may well be 'No one and not applicable.'

I've said this more obliquely before, but given a considerable amount of conversation and watching this firm behave, I'm pretty confident that the management doesn't care, doesn't understand, or is not acting toward his rational self-interest.

Here's a puzzler for some future business management class at Stanford: How do you take a protoype service with huge pent-up demand, people dying to pay money to you if only you can make it work half as good as intended, tons of press, evangalistic viral marketing done by your customers, all based around a few racks of servers and a really dumb graphical client and make it fail? It would be a shame if the correct answer to that exercise winds up being "ask Philip Rosedale to run the show". This is simply an amusing observation and is not a causal argument, but from a brief look at Real Networks, Inc. (RNWK) annual reports they have had only two profitable years: the ones immediately following Mr. Rosedale's departure.

Yeah.

It sure would be neat if a rather critical, show-stopping bug I discovered at the end of October would be fixed. One that I have been hotlining and IMing Lindens about for weeks.

Instead, we get ripple water! Yay!

'Rational' I'm afraid is the key word Mal...

Glad to see you clocked the Real link too. Did you ever try to buy a copy of Real Player during that period? Talk about make life difficult for someone willing to pay your firm money. I gave up. It was YEARS before I went back.
Woah! Deja vu or what!?

I just hope to hell the developers have got it sorted this time, otherwise I'm going to find a deep hole, climb inside, and pull the entrance in after me. Which is probably what Phil's 'plan B' is come to think of it.

They can't be fixing bugs. Fixing bugs is boring. Who the hell wants to spend 8 hours a day going over old code? BLOODY BOOORING!! If I was a programmer at Linden Lab and I was allowed to do what I wanted, then I'd be adding lots of cool features too!!!

I think the best managers are complete assholes. Obviously, the manager at Second Life is a very nice chap, so we're all just going have to suffer with lots of juicy bugs. Now imagine what Adolf Hitler could've done with Second Life. * shudders *.


Time for tea...

No, it's not a major update. It's adding two features and fixing a few bugs. Most people's pet bugs will not be among them, from the looks of things.

As Chris Linden explained, they've changed the release model, adopting more of a "when it's done" attitude to releasing features and changes developed by individual development teams. This way, they can get updates out the door faster, rather than having finished features sit and wait for three months while everyone else works on their own stuff.

So 1.8 shouldn't be thought of in the same way that 1.6 or 1.7 were, rather, consider it to be perhaps a 10th of those. It's only the numbering scheme that's changed here.

Also, when it comes to bugs, remember that not all developers are created equal! A developer that can change the UI around, or write an LSL feature is not necessarily up to fixing networking problems or the asset server. Seeing new features cropping up is not necessarily the result of the Lindens having misplaced priorities, but rather, using developer time where it can do the most good.

I appreciate your comments, and of course its a good thing to get the new features and fixes out the door as soon as possible, but does it really make sense to change the numbering system *away* from the way that the rest of the world used the point release numbering system? Why call it 1.8, when its really more like 1.7.2?

At this rate we will be using SecondLife 15.3 next Christmas.

I don't know much about technical this and that.

But I do know that the last time I could see SL properly, or move around in it very well, was in 1.6.

So I guess it's just gonna be this way forever now.

coco

I was going to reply to Doc with roughly the same reply as yours Cathy.. Oops, Catherine!!! I was about to hit that "Submit Reply" button but then I stopped to re-read Doc's post again. She's not complaining about weekly updates which is what I first thought. I love the idea of weekly updates and I imagine lots of other people do too, including Doc. I think she's asking what happens to all the bugs that are already present? Isn't it silly to start getting ready to release lots of updates with new features when there are so many bugs already present?

Isn't all the new features just going to add even more bugs to the ones already present? Isn't this going to be the end of the world as we know it? Aren't we all gonig to die? Will we ever see the sun again? Is Elvis really dead? Does anybody care?


Yours seriously, Starax

That reminds me Starax - we need to talk about a dolphin...

What happens when they get to Second Life 1.9 and need to release another update?

Surely not SL 2.0? That suggests a total rewrite.

It goes to 1.10

/me sticks a fork in Starax's eye.

I think what they're doing here is just releasing features that they would otherwise be working on anyway alongside bugfixes, rather than waiting. This isn't a huge change, and there will be bugfixes in 1.8.0, they'll just be the bugfixes we would have expected to see next week anyway, with a couple extra features.

It seems like what they're doing is switching to 1.8 to differentiate from the old numbering scheme. So we'll still get to 1.8.11 or whatever, but that might include new building features as well as the usual minor updates. Either that, or they'll go to 1.10 and so on. That could be confusing to some people though... like... uh... RESIDENTS.

they'll release 1.10. Software version numbers are not floating-point decimals.

I believe they're billing this as 1.8 to indicate the implementation point of the new development model. From here we'll probably see normally refined numbering as features are added weekly or biweekly (well a sprite can dream can't he).

If 2.0 really does have a fully implemented new rendering engine, mono, a client API, inbound RPC and Havok2, we'll likely have seen at least version 1.9.618.35 before it's released. How many ful client downloads is that?

New dev model doesn't make sense taking out the grid for 6 hours.

Having a system down is going to confuse the heck out of new users who then go around writing and saying bad things about SL.

The people at LL need to understand the nature of quality and how important it is when running a business.

Getting 90% of the things right doesn't work, because that last 10% kills you.

The major "bugs" are not so much bugs as design limitations and cannot be fixed without a major overhaul. Even though LL hasn't announced that SL1 is in maintainence mode and their principal developers are working on SL2, that is what is happening.

The SL1 crew has the unenviable task of keeping the service from failing under load more than 10 times the design limit. They are therefore trying to balance out various loads on the asset system - essentially robbing Peter to pay Paul - that is, if you want less lag from AVs you'll have to pay for it in script performance or texture performance.

This is also why we are seeing "features" like ripple water, because someone at LL believes that they can't have a "performance only" release. The P2P teleport was most decidedly a performance touch-up as it prevents clients from having to traverse up to 1km of content (assets) that they aren't really interested in. This should reduce asset server load significantly.

The "feature-by-feature" release plan described recently kind of scares me. If SL1 were clean, modular code, they could probably pull it off. There's good reason to believe that SL1 is hacked-to-death spaghetti code which is very sensitive to "small" changes. As a case in point, llSitTarget is a good example: none of the 1.6/1.7 releases should have affected it but it is still quite broken. When the code is in this state, simple changes need heavy regression testing which even the larger releases aren't getting.

So long as I'm wishing for a vPony, I only hope the SL2 team brought on a UI designer. The SL1 UI has always been an interface only an engineer can appreciate as the tasks are expressed in SL engine terms and not What-I-Want-to-Do terms. If you have any doubts of this, try making a gesture.

"It's not the failure, failure I can cope with; It's the hope I can't stand" - Withnail

Interesting point Blaze.

Even with rolling server updates and 'auto on logon' client updates (which nearly always freeze at about 80% for me), high frequency updates are going to result in unacceptable interuption of service.

The answer is definitely a vast improvement in quality control, coupled with a far more focused approach to choosing development goals.

That's the only way we are going to see relatively infrequent update downtime.

Haviing said that, I think a weekly update on the same day, at the same time is acceptable - indeed I believe other online VWs carry out regular updates in exactly that way?

It may or may not be Rosedale. It may just be too much idealism and not enough sense.

Hmm. Ok. It may be Rosedale.

Personally, I'm annoyed and offended at the wasted potential. But I am even MORE annoyed and offended that it is being wasted through stupidity and ignorance. Both are so god damn avoidable.

I am new to SL and one of the freebies so far. I tried it out since it sounded like a good idea..why would I want to kill dragons?

However I have some problems that I thought were just me. Your discussion helps me on at least two of them. The interface is too slow and my guy (avatar) can be a pain to control. Also, even with 1G caching set up my connection stops downloading after a minute or so even when I go to a new place (when I'm standing still). My connection isn't that fast (250-500K DSL)...so why isn't the program caching the stuff all around me? I am willing to wait & have plenty of disk space.

So since I have to set my draw distance way down to move comfortably, I get bored pretty quickly looking at gray areas and big squares.

Looks like some really talented people building stuff, but the economy appears broken (250L$ = 1US) with items going for $1US.

So I might visit once in a while but not real often. To bad it seemed like a good idea..maybe in another year or two.

Maybe this will help you folks work with SL..they are losing customers. At least one anyway.

eesh. WElcome to the forums, you shoudln't have come here.

The forums are bastions for unbridled negativity. Everything that may or may not be wrong with SL is posted here over and over until it is regarded as unbiased fact. It's the Fox News Channel for SL.

Responding to your points, if you're still here...

The economy is actually remarkably stable; a lopsided exchange rate at first glance has actually remained fairly stable around $L250/$US 1.00 for almost two years now. Monetary transactions keep increasing, so I guess SL's GDP is increasing too. Items going for $L250 are either low-quality or made by people who don't play the economy game. There are many items in SL that routinely go for thousands of Linden dollars a piece.

The cache problem is a bug, one that's allegedly being worked on. Actually, a while ago one of the devs posted on here that the reason we don't have unlimited cache sizes is becuase it would actually be *slower* than if it was working properly. Makes sense, I guess; the client would have to check and re-check every prim to make sure it hasn't changed since the last time you visited. Now imagine slogging through 10 gigs of such data every time you're flying around. Scary!

Anyways, SL is always changing. Come back in a few months and marvel at its difference.

LF

*Hands you a MMORPG making kit*

Here ya go, make it better.