WolfBaginski Bearsfoot

Might it not be that the weekly release of new code is too frequent? If there is such a tight schedule, there seems to be no margin for error. Is that wise?

But you have, in the past, been able to quickly revert to an earlier, safe, version of the code, which bypasses the whole QA element. Why was that choice ruled out? Somebody chose to keep the system running with broken code.

So you have fixed the ghosting issue. Great. But you're not deploying the fix for another 30+ hours. Not Great. If you do have an explanation for continuing to run server code which you know is broken, I would like to hear it. The best answer would be that you are running the fix through QA processes. That still doesn't explain why, when some regions are inaccessible because of the number of ghosted AVs, you have not reverted them to an earlier, known-safe, version of the code. It's Tuesday now, there has been time for a decision to be made at a senior level. There has been some good work done since the problem became obvious of Saturday. But the current situation leaves a distasteful impression that senior management at Linden Labs don't give a **bleep**.

I have, on many occasions, while travelling long distances with in-world vehicles, struggled to discern the locations of the boundaries between regions, which I know should be crossed at as low a speed as possible. Neither the nap not the minimap is as clear as it could be. May I praise the new solution which had been apparent this weekend, which shows frequently-crossed region boundaries as a clear row of dots on the map. While this new feature has not yet been documented (so, nothing new there), I infer that the more troublesome boundaries are more heavily marked. This is all a welcome improvement. The only, rather obvious, improvement I would suggest is a change in the colour of the marker dots, so that they can be distinguished from the dots marking the location of Avatars.

I suppose we all know that restarting the sim server is one of the common last-resort fixes for problems in SL, and we also expect this to happen at least once a week, as a side effect of the regular roll-outs of new code. From my point of view, not privy to the vast amount of information about what else, besides the sim servers, is getting fixed, fettled, and fscked by the Lindens, it looks as though the extended time between sim restarts, over the last six weeks or so, has greatly increased the problems of SL. In particular, while a sim that is routinely occupied, mainland or private island, has a chance of a restart being requested, my experience of the growing number of dreadful sim crossings is that areas of open water, all "Protected Land", may not be adequately monitored. My hypothesis is that the routine restarts have been hiding some serious underlying problems in the sim code. While it may not be formally classed as a memory leak, the gross symptoms are similar, in that at some point the accumulated effects slow down the sim. With a memory leak, the proximate cause of the slowdown may be a need to use virtual memory. I suppose you could have similar effects from a badly-coded logfile system, but I would have thought programmers have got past that carelessness long ago. The viewer code is, at least, able to be studied by outsiders, and hence we know that it has had problems. Some problems, over the last six months, are consistent with a "big picture" failure which misses warning signs, but people outside Linden Labs were able to spot the causes pretty quickly. The sim server code doesn't have that advantage. What's the fix for that? A project in the style of Viewer 2, to re-write the code from scratch? If so, all I can say is that I hope you have a solid specification od what the server code does.

These things are sometimes hard to pin down. I was getting messages about script errors--something about too many calls for an audio file--yesterday, every time I crossed a sim boundary. Your guess is as good as mine whether it's the script itself, viewer to sim-server, or something involving the inventory server. But it sometimes seems that Oskar is the only responsible Linden who takes the trouble to tell us things, and that doesn't help.

Will the MC sims get a restart?

I don't have the latest high-speed graphics card, but I have a suspicion that SL bottlenecks elsewhere than the graphics processor. The speed increases charted in the Wiki, at the upper end, are smaller than the GPU speed increases revealed by benchmarking. My experience is that RAM minimums for Windows and Windows software are claimed as ridiculously low. They always have been. For something such as a word processor in an office the performance might be usable, but no way it is letting the rest of the computer work well.

For everyone not using Internet Explorer, anyway,

Oskar, don't do that.

SL doesn't have any game servers in San Francisco, so you have been misled. I think the appropriate check locations are Phoenix and Dallas. You are getting a slightly worse ping time than I would expect from here in the UK for SL. My DL and UL speeds are limited by the ADSL line. I'd have to pay a lot more to get what you get. 1GB for SL? Out of how much system RAM? I am usually seeing more RAM used by SL, and I do wonder if the machine is a bit short of RAM. If there is disk-swapping going on, you are going to lose speed. Windows version is also a factor here, I don't know about other operating systems. If your video card is good enough to handle high frame-rate, and the bottlenecks are elsewhere, you can see minimal drops in frame rate as you increase the graphics quality setting, while the figure for ktris/sec increases. It can also take a few minutes for the frame rate to stabilise after you log in.

Since there is a new version of the Linden Lab viewer out, there are two possibilities. 1: It's been installed at an auto-upgrade, and, at least on your system, it has messed you up, big time. 2: It isn't installed, and it has the fixes you need. Firestorm, I'm afraid, is an old enough release to lag on a bunch of fixes that have been done. I will say that Cool VL Viewer from Henri Beauchamp seems to be a lot faster. He has been doing code optimisations rather than focusing just on the graphics. It's not a viewer for everyone, a bit too different in other ways, but it is different enough to be worth trying. If you still get the slow FPS, you likely will have to start looking at other details.

Firestorm has a different depiction of Ban Lines. I am not sure it is an improvement. Why do I say that? Well, with an SL vehicle, you are usually moviung faster than an unaided avatar, and the camera position is behind the vehicle. So the relatively short visibily range of a Ban Line becomes inadequate . If I could select some options, such as colour, boldness, and range, so that I can use discreet Ban Lines, or see them from half a region away, that would be a good thing. I don't know if there is any practical way to do this without a major code re-write, but there is a Ban Line HUD which gives some warning for a whole region.

I've recently observed behaviour which suggests that the bottlenecks on my system are not the graphics card. The frame rate barely changes with quality level, but the ktris/sec figure can see a six-fold increase between low and ultra. I've even seen fream rate increase when I've turned on shadows and DoF, which seems weird. There's a page in the Wiki which compares the SL performance of various graphics cards, and there are benchmark results elsenet. There hasn't been the increase in SL performance to match the increase in GPU performance. Henri Beauchamp claims that there are some significant optimisations that can be made to the code. I've tried his Cool VL Viewer, It does give noticeably higher frame rates. I can't help wondering if, when Viewer 2 was outsourced, Linden Lab started losing the institutional knowledge needed to maintain their Viewer. But the TPVs start from the same codebase. Maybe their developers, spending time in SL for fun, know better what needs to be fixed.

There's some big differences, such as SL needing to download graphics data over the net. And I can see the overall system, server code running the regions and the network connecting them, as being well behind the curve because of early decisions made. It wouldn't surprise me if 512m regions would be a viable choice now, but converting the whole grid would be difficult. Researchers have figured out how to run sims with a hundred active residents. I doubt Linden Lab can afford to start from scratch on their code.

I've just had a Viewer crash, with an arcane error message on-screen. Couldn't copy-and-paste the text. Will it be in a logfile somewhere? Is it even worth trying to tell the Lindens?

I don't use the Linden Lab viewer myself, and there have been a couple of times when reports have suggested that the upgrade was not a success, in terms of performance. There's some sense to automatic upgrades. It doesn't excuse the poor communication, such as has masked the big changes of the last few months. I have, a few times, experienced automatic upgrades on other software which have been a disaster. It's something which has to be done right, because of the software supplier is effectively being given permission to manipulate a customer's computer. It's one of those potentially expensive legal problems if something goes wrong: what permissions did you ask for, and what did you assume without good cause? Telling a customer that an upgrade is out, even nagging them about it is a lot safer, legally, when things go wrong.

Most of what you're looking for is in Firestorm now, better than 90%. And, yeah, the UI is different enough from Phoenix to be awkward, but it isn't the hell-spawned abortion that is Viewer 2 or the current Viewer 3. To me, one of the essential features is the ability to change the colour scheme so I can read the chat and IM. The default color scheme in the Linden Viewer is unusable for me, I need to download a third-party add-on just the get readable text, and a different version has to be downloaded for every released version of the Linden Viewer.. The Linden Viewer has been like that for close on two years. Most of the third party viewers have fixes built in. Sure, they all depend on the Linden Lab code, but they fix the problem which Linden Labs doesn't.

My own experience suggests that my particular bottleneck is not the graphics card, Evidence? Open the Stats windows (Ctrl-Shift-1) and look at the Advanced | Render section. The triangle is the basic unit of graphics, and the ktris drawn per second gives yoiu an indication of what the graphics card is doing. What I have found is that increasing the Graphics quality in Preferences has a small effect on frame rate, and greatly increases the ktris drawn. This seems to be independent of the Viewer used, though specific numbers do differ. I've read that Henri Beaumont has done some significant code optimisations to his Cool VL Viewer, and it does seem to perform well, but that is a mix of v1 and v2+ code, to get Mesh into a v1-style Viewer. |I may have to upgrade to Windows 7, so as to get the best out of multi-core processors, but i am not sure if the Viewer can take advantage of it. Speculating wildly, I suspect a part of my problem is how the network interface works. Is the code in the Viewer which handles incoming network data competing with the code preparing graphics data for the graphics card? What I see is consistent with that. It doesn't help that a lot of ISPs have over-sold capacity. When they set up their businesses, people didn't use always-on connections to stream media. At peak times, your share of their network capacity can be much less than the advertised figure. Which is one reason why the Viewer's bandwidth setting might be usefully reduced. It's a tricky figure to interpret, but the Stats Window tells you the instantaneous packet loss, and if packet loss coincides with bursts of high bandwidth usage, it's likely worth reducing the bandwidth setting. OK, nothing here about what's good hardware, but some clues about finding what can be done, I hope. I wonder if out-sourcing Viewer 2 lost the Linden expertise to deal with such things?

There have always been problems with clothing and the user's body shape. You can see things such as doglegs in bikini straps as you shift the breast size/position settings. Sculpts and prims need careful adjustment. Qarl's fix is welcome. Being able to save and add a head-only shape would have avoided a lot of the problems I have had. It's not quite a universal, but the head and face are surely far more significant an element of personal identity than a body-shape. Even for a Furry, a different set of head data can shift the attachment points for head-components such as the jaw and eyes. It is slightly astonishing that this has never been noticed.

Soþlice on þam dagum wæs geworden gebod fram þam casere Augusto, þæt eall ymbehwyrft wære tomearcod. Þeos tomearcodnes wæs æryst geworden fram þam deman Syrige Cirino. And ealle hig eodon, and syndrige ferdon on hyra ceastre. Ða ferde Iosep fram Galilea of þære ceastre Nazareth on Iudeisce ceastre Dauides, seo is genemned Beþleem, for þam þe he wæs of Dauides huse and hirede; þæt he ferde mid Marian þe him beweddod wæs, and wæs geeacnod. Soþlice wæs geworden þa hi þar wæron, hire dagas wæron gefyllede þæt heo cende. And heo cende hyre frumcennedan sunu, and hine mid cildclaþum bewand, and hine on binne alede, for þam þe hig næfdon rum on cumena huse. And hyrdas wæron on þam ylcan rice waciende, and nihtwæccan healdende ofer heora heorda. Þa stod Drihtnes engel wiþ hig, and Godes beorhtnes him ymbe scean; and hi him mycelum ege adredon. And se engel him to cwæð, Nelle ge eow adrædan; soþlice nu ic eow bodie mycelne gefean, se bið eallum folce; for þam to dæg eow ys Hælend acenned, se is Drihten Crist, on Dauides ceastre. And þis tacen eow byð: Ge gemetað an cild hræglum bewunden, and on binne aled. And þa wæs færinga geworden mid þam engle mycelnes heofenlices werydes, God heriendra and þus cweþendra, Gode sy wuldor on heahnesse, and on eorðan sybb mannum godes willan.

About the only thing the Marketplace is good for is as a search engine, and it's not so good at that. Non-delivery has become so frequent a problem for me that I feel it is just not worth buying anything through the Marketplace (this problem may be exacerbated by the UI design inbtroduced in Viewer 2). In-world sales and delivery can sometimes be slow, but they work. The choice seems easy.

It's no good. You keep saying things like this, and all my years of education keep nagging at me. [Posted 8:26am PST, 21 December 2011] We are beginning this morning’s maintenance on the AGNI main grid, which will require an extended period of region restarts. We expect this maintenance to be completed by December 25, 2011. We will update this when work begins, as well as throughout the process. As my old English Teacher was wont to sharply enquire, which "this maintenance" is expected to be completed when, and why "We are beginning" and "We will update this when work begins" There are kids leaving British Primary Schools who can write better than that.

InWorldz is an OpenSim-based shared world. Smaller than SL, but using the same basic tech. So it shouldn't be lumped in with the other MMORPGs. It may help to reduce the bandwidth setting in the Viewer, but since you're using a laptop I suppose you might be using Wi-Fi. If you are, your connection could be getting strangled by the effects of users on the same channel as you. Some wi-fi software only needs the channel to be set on the wi-fi router, other stuff can need an adjustment on the computer as well. I recommend setting this up over an ethernet connection to the router, rather than adjusting wi-fi settings over the wi-fi link.

As a Marketplace merchant, on a very small scale, LL takes a fee from me for every delivery. That sets up all the requirements for a legal contract to exist between me and them to deliver the goods to the purchaser. And they're failing. It doesn't matter whether the L$ is legally money or not. There's an offer and acceptance between me and LL when I put stuff on sale through the Marketplace, and every sale involves considerations, between the purchaser and me, and between me and LL. It's gets silly because of the size of the transactions. How many microseconds of lawyer-time can you get for 1 L$? But LL are being paid to deliver goods, and they are failing to fulfil that contract. It isn't a dispute between residents we're talking about, and, if there is a get-out clause in the TOS it may be unenforceable, as contradictory to statute law. I'll let somebody else take them to court. But it's possible that the fee is opening up a few legal cans of worms in other areas.