SegmentationFault String

Several times today I have gotten the message: "We are sorry, Linden Lab has discovered degraded performance on your connection to the region you are on. You will need to restart Second Life and log into a new region for the next 30 minutes to an hour. We apologize for the inconvenience." What conditions trigger that message? My system is reasonably fast, with 16GB RAM and a mid-range gaming graphics card; according to speedtest.net I'm getting about 4.9Mbps throughput (for a rural connection, that's amazingly good). As shown in the attached screen shot my fps rate is excellent (the 323 is misleading since I didn't get this until after SL stopped rendering anything; normally it's between 35 and 90). There have been no changes to my connection, my drivers, or my hardware since yesterday. I tried clearing my cache but that did not help. The message came up on relatively unoccupied sims, not heavily crowded areas.  NOTE: I am aware this is identical to JIRA issue #VWR-25566 but after some five years that still has no answers, just more people reporting the same message. If I at least knew what factors cause it to display that message I would have a better idea where to diagnose potential problems on my end, but as it is the message is completely useless in terms of troubleshooting. I should have known when I chose my screen name that it would be all too prophetic...

It was a timing issue. To prevent exceeding the limit on llHTTPRequest I follow every HTTP request with llSleep of 0.9 seconds. Since I split HTTP off into a child script, requests from the parent script were getting ignored while the child script was in llSleep. I had to make the parent script smart enough to know not to keep sending requests until the child script was ready for them, and the child script smart enough to only do llSleep if it detected too many llHTTPRequest calls in too short a time.

I have a very large script that was running out of memory so I broke it into two scripts, moving HTTP communication and data lists off to a child script. The main script uses llMessageLinked to pass requests on to the child script, like so: llMessageLinked(LINK_THIS, _COMMANDS, sMyRequest, gkOwner) // where gkOwner is a global previously declared as llGetOwner() I put an llOwnerSay in the main script when it first sends the request and in the child script at the start of link_message, and got these results back: [07:44:23] Prototype: main sending pickcontact over _COMMANDS [07:44:45] Prototype: child got command over _COMMANDS channel I have reset the scripts, detached and reattached the object, tried LINK_SET instead of LINK_THIS, moved to different regions, left it overnight and restarted in the morning, and in every instance the lag ruins timing. It was actually worse with LINK_SET, over 70 seconds. The only time either script sleeps is when the child script issues an llHTTPRequest, at which point it sleeps for 1.0 seconds to avoid exceeding the HTTP cap. But in this case, no HTTP requests were posted; the child script was just sitting idle waiting for the link_message event to trigger. Even weirder, it's *only* this one event. Other link messages go through in under 1 second, exactly as they should. Has anyone seen this before? Any suggestions on how to fix it? I have been scripting for six years, and have used llMessageLink thousands of times... never saw it delay like this before.

Hah! Thanks, Theresa. Glad to know it's not just me. Jewel and Marigold, after all these years I'm very familiar with the amazing skins available both for free and for reasonable proces in the marketplace. But as I said, I'm fond of my self-designed avatar with her nerdy, totally non-glamorous freckle face and thick glasses. Obviously I'll have to settle for somebody else's design since that feature no longer works properly, but it's a shame. It's also a moot point. Since my RL wife started joining me in SL, I almost always use my male avatar. She feels weird going out dancing with the homely nerd girl, and the male almost never wears any kind of makeup :-)

When I started playing Second Life five years ago, I had no money and did everything on the cheap. Rather than buy a photorealistic skin, I designed my own avatar using the system appearance editor. Even after I could afford any skin I wanted, I had grown fond of my quirky appearance and kept it. So my female avatar doesn't have some glamorous screen star skin. Instead, I used the appearance editor -> Face Details option to add freckles, rosy cheeks, and light pink lips. About a year ago, something went wrong and she looked like a goth girl: Those innocent flushed cheeks now are black, as are the lips. The freckles still look OK, but they're dark brown to begin with so who can tell? It's not my viewer: In recent weeks I have tried the latest LL Viewer, Cool VL Viewer, and Singularity viewer on both a Linux machine and a Dell running Windows 7 (as well as a VMWare virtual Windows 8 session on the Dell). In preferences->Graphics, I have all shaders enabled except water reflections; local lights, hardware skinning, and avatar cloth are enabled. Draw distance is 128, Post Process Quality high, and all mesh detail options are on high except trees, terrain, and sky. Changing any of those has no effect on the goth black face detail. I keep my sun position on high noon all the time and I have even put a big honkin' face light right in front of me that only washes out all detail while leaving cheeks and lips black. Interesting thing about the LL example on the left: You can see the original intended effect of rosy cheeks on one side of the face, but not on the other. When exited out, ran the other two viewers, then went back to the LL viewer for a second shot, that redness was gone and it looked exactly like the others, black on both sides. I tried facing different directions and changing sun to morning and evening, but I never got that rosy glow back. As a workaround I can leave off the cheek and lip details and apply a tattoo layer, but that means I have *zero* control over color and intensity, which makes it an annoying workaround. Has anyone else experienced this, or am I the only one on the grid who still uses system appearances?

I was working on a furry AO that involves replacing the standard standing & walking animations with a position on hands and knees for crawling. Learned qavimator, created a fairly smooth crawl... but there's a problem with the transition from standing (on hands and knees) to walking (crawling). For a brief instant, the avatar goes from horizontal to vertical, while still retaining the bent arms and legs of the kneeling position, before going back down to the right position to begin crawling during the walk phase. Changing back from walk to crawl doesn't seem to do this. Per the qavimator instructions I left frame 1 alone, in the standing position, and told it to start looping at frame 2. Just for fun I tried replacing frame 1 with another kneeling position, but the result was that the avi walks around upright, but with arms and legs bend in the crawl position. The whole reason I started this project was to replace a freebie animation that has the same problem, so it's not just me. What causes this, and is there a way around it? The whole illusion is ruined when the avi suddenly jumps upright before starting to crawl. Contact me inworld if you need to see what I'm talking about, or if you need me to send you the .bvh file.

By "get a photo" I mean with my digital camera, which in turn means I can get whatever angle works best. But from what you're telling me, that's not going to do a lot of good. Sigh. Thanks for explaining the limitations of the art so I can understand, anyway. Extra points for quoting from "Real Genius" :-)

I'm getting fairly comfortable with the Chip Midnight / Robin Sojourner templates so I can make shirts and pants fairly easily, but shoes are still a complete mystery. Does anyone have any experience using the templates to design non-prim shoes? Say I have a pair of shoes I love so much that I want to put them on my avatar. I get a picture of them, pull up the photo and the lower body template in Gimp, and... there's just no way to match up the shape of the actual photo with the shape of the template. Toes that don't even come to much of a point are stretched out of proportion on the template, there's no real clue where top surface ends and sides begin... If I could see an example PSD with the shoe texture filled in, I could probably understand how to do the same with my own shoes.

I never got into vi... too dang many key combinations to remember. Its only advantage is that it's truly hardware independent and doesn't require a GUI. But for most of my code editing in any language on the linux box, I'm all gedit now. It even has syntax highlighting specifically for lsl now! I haven't really found a good syntax highlighter for Winduhs yet. For work we have to use Visual Studio which is about as bloated as you can get and only supports languages that MS produces. So for anything else, I work on my Linux box in gedit and copy it over to the XP box.

I wish The Powers That Be would allow llInstantMessage to work between rezzed objects. That would solve *all* inter-sim problems.

Thanks for the fast reply, Cerise. I understand about the Mozilla thing, but what threw me off was that it *just* stopped working today. Up until now, I think SL's default user-agent string included the word "Mozilla". But a quick test from my server reveals that it now sends: Second Life LSL/11.03.08.223047 (http://secondlife.com) ... unless you use the hack that you describe. Interestingly, I had exactly the same problem with OpenSim only in that case the hack doesn't work because the appended user agent string is stripped off before the URL is submitted.

Up until yesterday, I was making great progress on a script that pulls shoutcast stream data from my shoutcast server using: llHTTPRequest("http://my.shoutcast.server:8080/7.html",[],""); Now suddenly it gets a timeout error. I can still get that data from an external browser so I know the server on my end has not changed. And in lsl I can still use llHTTPRequest on any address that's a regular port 80. But any other port... not so much. Has something changed in the way secondlife handles http requests that no longer accepts data from nonstandard ports? If this is so, it pretty much kills every DJ venue on the grid that relies on that ability to display song titles. And if I'm mistaken, tell me I'm an idiot and please show me what I'm doing wrong. Thanks!

I just finished a set of scripts for a HUD that manages shoutcast stream info for me. This is the mother of all shoutcast stream utilities -- it does EVERYTHING. Because I use different streams depending on which venue I work, it lets me maintain a list of multiple streams and toggle which one(s) to fetch info from. Because some venues let me set up my own tipjar and some provide one, and some venues already have a song title display but others I need to provide it, my HUD will send the song data to any number of different devices I keep around -- you could also have it send the song data to one of those big scrolling marquees, but I have not tried that. But wait, there's more... I have some friends who listen to the show in RL, so for them I set up a web page where they can post song requests. My HUD will pick up those requests and tell me when a new one comes in, plus let me delete them as I play them. This sucker weighs in at almost 1100 lines of code and more than 50K of script memory. It has more bells and whistles than Times Square on New Year's Eve. Anybody interested in trying it out? It seems a shame to not share it with others, but I have a feeling it may be too complicated for non-programmers so I could use another DJ's opinion of it.

As a DJ and a scripting geek I have played around with various tools to help me collect data from the various streams I use and display the information (song titles, listener counts, etc.). Plus, I have some RL friends who listen to the streams directly and they asked for a way to submit song requests outside of SL. So I put all this functionality into one big honkin' HUD that will poll multiple stream servers (because I never know which one I'll be using depending on the venue), fetch song requests from my website request form, and display whatever info I want as floating text above tipjars, or (if the venue uses its own tipjars) some other device. I like it because it literally does EVERYTHING for me, and I'd love to share it with other DJs... but I have a feeling it's way too complicated for non-geeks. Anyone out there interested in giving it a try and telling me what you think? IM me in-world, or email me segfault@steigerfamily.com.