Chat log from the meeting on 2017-08-29

From OpenSimulator

Revision as of 13:30, 29 August 2017 by Sheera Khan (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

[11:09] Arielle.Popstar whats happening in the world of opensim this week?
[11:09] Selby.Evans
[11:10] Arielle.Popstar mm yes saw that
[11:10] Gavin.Hird Do you really want to discuss that here Selby?
[11:10] Andrew.Hellershanks Not that much new since the latest release.
[11:10] Gavin.Hird Perhaps you should discuss it with the person you copied to code from?
[11:10] Arielle.Popstar yes lets give that a wide berth :)
[11:11] James.atLLOUD /me is viewing in Kokua today
[11:11] Selby.Evans Gavin - not mentioned for discussion. -- just what's happening
[11:11] Steve.Franklin Kokua works really well for me
[11:11] Steve.Franklin On Mac...
[11:12] Gavin.Hird Sounds like you are proud of announcing your own theft
[11:12] Leighton.Marjoram personally I have little success with that viewer and stick with Firestorm
[11:12] Andrew.Hellershanks I rarely use Kokua but I do have it on my machine along with about 5 other viewers.
[11:12] Gavin.Hird Glad to hear Steve
[11:12] Steve.Franklin I had no part in that
[11:12] James.atLLOUD I'm hooked on phototools Leighton so will always use firestorm for those.
[11:12] Bill.Blight I use Kokua quite a bit, just a few interface things I like better in FS,
[11:12] Selby.Evans I am mainly interested in OS dedicated viewer.
[11:12] OtakuMegane Desu: Kokua works for me on Mac but is slow compared to Firestorm.
[11:13] Gavin.Hird Try my latest builds from here
[11:13] Steve.Franklin I tend to use Singularity as I am used to it, but I am exploring...
[11:13] Arielle.Popstar i am leery of a opensim dedicated viewer
[11:13] Gavin.Hird they are signed and have much more recent code than the ones NickyP has built
[11:13] Arielle.Popstar i find too often that viewers dont all work for everyone
[11:13] Steve.Franklin I hardly ever go to SL anymore
[11:14] Gavin.Hird
[11:14] Gavin.Hird these are macOs only
[11:15] Andrew.Hellershanks James, phototools?
[11:15] Steve.Franklin Yes I noticed that Gavin and I am really glad you are a Mac guy!
[11:15] Steve.Franklin lol
[11:15] Gavin.Hird :-))
[11:15] Steve.Franklin warms the cockles of my heart!
[11:15] Gavin.Hird There are very fresh builds of the SL version too
[11:15] Arielle.Popstar i think the effort should be expended in making opensim compatible with all known viewers rather then focusing on only one
[11:15] OtakuMegane Desu: Will check those out. Actually my biggest issue right now is the upgrade to Sierra broke a bunch of the advanced rendering stuff.
[11:15] Gavin.Hird
[11:16] Gavin.Hird meaning Arielle?
[11:16] Steve.Franklin yes I am still running 10.10.5
[11:16] James.atLLOUD @andrew - phototools is a panel in firestorm with dedicated camera and environment controls. just super handy.
[11:16] Gavin.Hird the new builds will only run on 10.11 or newer
[11:17] Bill.Blight YOu can't do that arielle , You can't have the server software chasing the viewers
[11:17] Steve.Franklin ah then time for a little upgrade! lol
[11:17] Andrew.Hellershanks James, ok. I don't use FS that often. I've not looked at it closely enough to see what unique features it has.
[11:17] Arielle.Popstar meaning that if we start focusing on only one viewer, the day will come that half the people will not be able to log in, install, or hav issues with that one viewer
[11:18] Gavin.Hird right now just about every viewer out there has sole focus on SL with OpenSim as an afterthought
[11:18] Leighton.Marjoram I like the choice of viewers, I mainly use Firestorm and Alchemy.
[11:18] Steve.Franklin I think having a bunch of viewers is a very good thing
[11:18] OtakuMegane Desu: The problem with that is Opensim will either have to get any new features enabled in every viewer or forever confine itself to what SL offers.
[11:18] Arielle.Popstar they all have an opensim compatible one
[11:18] Gavin.Hird the KokuaOS code base is the only one that has diverged from that
[11:18] Andrew.Hellershanks The idea of a dedicated OS viewer has been talked about before. If there was an OS specific viewer that wasn't tied to the SL viewer it would allow changes to the OS code based not currently possible.
[11:18] Freaky.Tech you guys know what the real key to success is? communication in an objective way
[11:19] Arielle.Popstar Otaku, what would opensim offer if it diverges from s/l?
[11:19] Steve.Franklin yes
[11:19] James.atLLOUD that's what I understood @Andrew dedicated viewer
[11:19] Arielle.Popstar at this point s/l sets the standards and creates the motivation to make opensim keep up
[11:19] Gavin.Hird Andrew there is nothing preventing you or anyone from starting to draft the changes and protocols
[11:19] Bill.Blight What couldn't opensimulator offer if it was cut off completely form SL, the sky is the limit
[11:19] Gavin.Hird they can be worked into any viewer
[11:19] OtakuMegane Desu: Right now, not a whole lot. In part because many possible significant changes would require viewer changes as well. And that's hard enough as it is.
[11:19] Freaky.Tech you know why viewers in SL work well because Linden Labs communicates with the viewer people
[11:19] Arielle.Popstar we have one active core dev
[11:19] Steve.Franklin Yes in some ways SL holds back OS
[11:20] Gavin.Hird question is what do you want to change, because every time someone asks, there is virtually no response but some piddly little things
[11:20] Andrew.Hellershanks Gavin, It would have to be someone else as I don't know enough of the underlying details of the communications protocols between viewer and Open Sim.
[11:20] Selby.Evans Freaky -- you posted something that suggested a correction needed in the viewers regarding the handling of mesh -- does that still stand?
[11:20] Gavin.Hird come with the specs of what you want
[11:20] Arielle.Popstar exactly Freaky whereas opensim core doesnt afaik
[11:20] Steve.Franklin I could give out my list! lol
[11:20] Gavin.Hird development does not start with the code
[11:20] Freaky.Tech got updated
[11:21] Gavin.Hird development starts with ideas prototyping specing it out
[11:21] Freaky.Tech and is related to a as I heard decided limitation on ubOde
[11:21] Gavin.Hird then you can start coding
[11:21] Steve.Franklin yes a plan or roadmap is a good place to start
[11:21] Andrew.Hellershanks Gavin, there are some issues with the protocols. I don't remember the specifics. IIRC, there are ways to improve the performance of OS by improving/changing some of the messages being passed between viewer and OS.
[11:21] Selby.Evans Thanks, Freaky
[11:22] Gavin.Hird ok, then get the specs out
[11:22] Gavin.Hird as you used to say, start making Mantises
[11:22] Gavin.Hird point out where the bottlenecks as seen from server is in the viewer
[11:22] Gavin.Hird tell us
[11:23] Gavin.Hird cause right now there is maybe - maybe one opensim specific cap
[11:24] Steve.Franklin Everything that is done with OS should be both server AND viewer as both are are intricately related
[11:24] Gavin.Hird exactly Steve
[11:24] Gavin.Hird but first it starts with ideas and what you want to achieve
[11:24] Leighton.Marjoram Nods @ Steve
[11:24] Andrew.Hellershanks steve, good in theory but it is two completely different code bases written in two different programming languages.
[11:25] Steve.Franklin That was a realization that I came to recently. We server people tend to ignore the viewer people- we should all be working together to bring OS into the future
[11:25] Bill.Blight Would be nice if Core could generate a roadmap to better performance, and let people know where the best routes lie with the current implementation .. And laying out possible future paths to better performance
[11:25] Arielle.Popstar there were a number of OS specific changes that rolled out to all the viewers in past so i dont see the issue
[11:25] Gavin.Hird Melanie has repeatedly refused to make a roadmap
[11:25] Bill.Blight yep, as I said, "would be nice"
[11:25] Gavin.Hird Diva is not exactly helpful too
[11:26] Steve.Franklin maybe we should just get a group of people together and start a roadmap of our own. Throw it out there and see who or what jumps on
[11:27] Gavin.Hird yeh
[11:27] Steve.Franklin I would be interested in being a part of that as me and ADB
[11:27] Arielle.Popstar sounds nice but unless you got some C# devs willing to work on it it means little until you have core onboard
[11:27] Freaky.Tech open source people do not work that way. A lot of projects have been defined that way.
[11:27] Arielle.Popstar Freaky the only one i know who has something
[11:28] Freaky.Tech of someone asking whether someone else would implement their project
[11:28] Steve.Franklin Don't you have a plan for what you are doing Freaky? Have you written it up somewhere?
[11:28] Freaky.Tech this is not my place here ;)
[11:29] Arielle.Popstar what would you do if you were here /?
[11:29] Steve.Franklin well true. I just think that we would make a lot more progress working together with goals in mind
[11:29] Freaky.Tech I simply wrote it: get the past to be the past and make dev people go communicate
[11:29] Steve.Franklin Sorry I am probably jumping in where I should not...
[11:30] Steve.Franklin yes freaky exactly
[11:30] Arielle.Popstar they did at one time
[11:30] Bill.Blight a road map does not have to be a world atlas,
[11:30] Leighton.Marjoram A few lines with time frames would be a start
[11:30] Andrew.Hellershanks steve, that is a good idea but some Open Source projects run without clear directions on changes.
[11:30] Steve.Franklin no not at all
[11:31] Steve.Franklin it can be very short at first!
[11:31] Steve.Franklin just a block or two!
[11:31] Arielle.Popstar personally i think opensim should carve out a niche for itself by concentrating on inworld building tools
[11:31] Freaky.Tech any open source project knows what points it wants to achieve
[11:31] Freaky.Tech deadlines are not necessary
[11:31] Andrew.Hellershanks Freaky, world domination! ;)
[11:31] Gavin.Hird exactly Freaky
[11:31] Arielle.Popstar all the other vw's are focused on blender and other mesh software
[11:32] Steve.Franklin did not say anything about deadlines Freaky- they aren't really needed
[11:32] Freaky.Tech let's say it that way you know there will be a next version whatever
[11:32] Gavin.Hird did Apache and Linux succeed without goals?
[11:32] Bill.Blight Deadlines lead to rushed code
[11:32] Freaky.Tech you will at least know what points you want to have in it
[11:33] Andrew.Hellershanks It is almost pointless to set deadlines on Open Source projects.
[11:33] Steve.Franklin yes
[11:33] Freaky.Tech deadlines with voluntary work does not work
[11:33] Steve.Franklin and the viewer and server map should be side by side
[11:33] Gavin.Hird Deadlines usually implies you have resources that can realistically reach the goals
[11:33] Freaky.Tech but knowing what will be in next version does
[11:33] Arielle.Popstar gives a guideline
[11:33] Gavin.Hird OpenSim does not have resources
[11:33] Leighton.Marjoram may be something like this
[11:33] Leighton.Marjoram
[11:33] Leighton.Marjoram That is the Ghost Blog CMS road map
[11:34] Steve.Franklin Freaky has it there- knowing what is going to be in the next version
[11:34] Selby.Evans You can't look too far ahead when you don't know what people will want
[11:34] Andrew.Hellershanks A list of goals (apart from fixing bugs and other items) in a project mantis can be quite useful.
[11:34] Bill.Blight well OS used to track better, requests, and future additions on the wiki, and we used to know what direction we were headed or where we are .. That has kind of fallen away
[11:35] Freaky.Tech Selby you will not know everything
[11:35] Freaky.Tech just the major goals
[11:35] Steve.Franklin yes exactly
[11:35] Arielle.Popstar because the maintainer is gone
[11:35] Selby.Evans Right, Freaky
[11:35] Arielle.Popstar Justin
[11:35] Steve.Franklin and some will change over time as tech advances
[11:35] Gavin.Hird Selby do you really thing Apple would be market leader with your attitude? Apple's focus is MARKET CREATION
[11:36] Gavin.Hird that should be the goal for OpenSim too
[11:36] Gavin.Hird create a market for soemthing people yet don't know they want
[11:36] Arielle.Popstar niche market
[11:36] Steve.Franklin yes we need to make OS the one everyone wants
[11:36] Bill.Blight You also can't focus a server based software solely on what the PEOPLE want ... Leave that up to third party people .. Opensimulator as a server platform was never really intended for "end user" use .. You get into a whole new can of worms and a mess when make that too much of a foucus
[11:36] Selby.Evans Agreed, Gavin.
[11:36] Steve.Franklin and not just becausage it is vintage!
[11:36] Arielle.Popstar bs Bill
[11:36] Gavin.Hird 500 million devices sold per year is a good niche Arielle
[11:36] Bill.Blight exactly what I think of your opinion
[11:37] Arielle.Popstar talking like Mel
[11:37] Steve.Franklin Apple has already created market rather than just being a me too
[11:37] Leighton.Marjoram @ Steve Vintage is the new Retro :)
[11:37] Steve.Franklin me too is more like Microsoft.
[11:37] Steve.Franklin well there is that Leighton!
[11:38] Steve.Franklin lol
[11:38] Gavin.Hird Bill Gates is the biggest thief of the lot
[11:38] Steve.Franklin well another me too
[11:38] Steve.Franklin Go to a CES show and see the new stuff and then the hundreds of cheap me too copies
[11:38] Bill.Blight yep
[11:38] Steve.Franklin it is fun to watch
[11:38] Bill.Blight and right now OS is a cheap me too copy
[11:39] Gavin.Hird both fun and pathetic
[11:39] Steve.Franklin that is why I search out the new things and go talk to them
[11:39] Steve.Franklin OS has a ton of potential that just needs to be thought out
[11:40] Gavin.Hird I agree Steve, but then we cannot have someone calling themselves core sitting there as nay-sayers
[11:40] Arielle Popstar: i think a lot is thought out but we dont have the development team to make it a reality
[11:40] Bill.Blight But as an example, Apple, who I don't like but is very successful, tells the users what the new features are, they say "THIS IS HOW IT IS" .
[11:41] Leighton.Marjoram They define the reality of their market
[11:41] Arielle Popstar: and thats why they losing market share
[11:41] Gavin.Hird They do Bill when they introduce a new product
[11:41] Gavin.Hird but not after
[11:41] Bill.Blight (usually bought or stolen from someone else though)
[11:41] Gavin.Hird the amount of feedback that goes back to them via both marketing channels and not the least the developers is staggering
[11:42] steve Franklin: Apple is also learning how to respond with things like Mac Pro. A whole lot of community input on that
[11:42] Gavin.Hird every WWDC they introduce a shitload of new functionality to enable developers to create better
[11:42] steve Franklin: yes
[11:43] Gavin.Hird they dont have 80% of the mobile space profit by not listening to users
[11:43] Gavin.Hird not possible
[11:43] steve Franklin: and they learn their lessons by listening to the people who buy their stuff
[11:43] steve Franklin: That is why there was a big iPhone too
[11:43] Bill.Blight the difference ... in my analogy is this ... Apple makes "end user" devices, OS is a server platform ... Do Server makers listen to desktop users when making server adjustments? No they listen to server admins ... OS was never meant to be an end user platform,
[11:44] steve Franklin: yes Apple is far from perfect but they are somewhat trainable
[11:44] Freaky.Tech you know where this kind of bashing all the time leads to => just having unnecessary rage
[11:44] steve Franklin: but the viewers are end user Bill and it all ties in together
[11:44] Gavin.Hird without end users opensim server can sit there burning CPU for itself
[11:45] Leighton.Marjoram agree Freaky and no action, just rage
[11:45] Gavin.Hird opensim does not have ANY purpose without end users
[11:45] Gavin.Hird none whatsoever
[11:45] Arielle Popstar: exactly
[11:45] Bill.Blight soooo with that thought steve, end users should help design web server software? Since their browser is end user client?
[11:45] Leighton.Marjoram ends users and server admins aren't necessarily the same person
[11:45] Arielle Popstar: it is end user ability to run it at home that gave it its start
[11:45] Gavin.Hird and the end user's ability to create
[11:46] Arielle Popstar: inworld
[11:46] Gavin.Hird the admin is just a means to the end goal of the end user
[11:46] steve Franklin: I agree with Gavin
[11:47] Gavin.Hird just like a cell phone grid would be completely pointless without end user devices
[11:47] steve Franklin: And with Freaky- We don't need any rage- Got enough of that in the US already!
[11:47] Arielle Popstar: early ops devs tried to make the server software user friendly for the end user
[11:47] James atLLOUD: Did anyone hear Melanie a couple of weeks ago on Mal's show say she had wanted to buy opensim. What would have happened then?
[11:47] Arielle Popstar: not just server admins
[11:47] Bill.Blight without end users there would be no need for web pages either, so we should just develop ALL web sites and web servers to the whim of the end user , got it ... I'll shut up now
[11:47] Andrew Hellershanks: A company, such as Apple, has a different motivation for working on software or hardware than people working on Open Source projects. Apple has lots of staff and can easily hire more to work on development.
[11:48] Andrew Hellershanks: Hard to do the same thing when a project is being worked on by people volunteering their time.
[11:48] steve Franklin: To Bill no but remember it all ties together the viewer and the server. If the viewer can do some new thing that the server does not support that it is all for naught
[11:48] Arielle Popstar: like pathfinding
[11:49] Arielle Popstar: hypergrid for s/l
[11:49] Gavin.Hird pathfinding is in the viewer code, but does it hurt you as the end user it is there?
[11:49] Arielle Popstar: nope
[11:49] Gavin.Hird no it does not
[11:49] Arielle Popstar: but at this point it doesnt benefit opensim
[11:49] steve Franklin: Bill you were part of who convinced me that I needed to look at viewers way more carefully
[11:49] Gavin.Hird for the opensim viewer it is pointless
[11:49] Gavin.Hird so it is on the plan to be removed for Kokua OS
[11:50] Arielle Popstar: maybe down the road it may be of value
[11:50] Gavin.Hird a lot of code is planned from being removed from the Kokua OS codebase, but it is not very easy
[11:50] Arielle Popstar: if a dev looks at making it work for opensim
[11:50] Gavin.Hird as it has been added over years, and the minut you start touching it other things fall over, also for opensim connections
[11:51] Gavin.Hird so removal is very hard actually
[11:51] steve Franklin: yes true- you have to keep track of the falling dominos!
[11:51] Arielle Popstar: so why not leave it in and that way down the road it doesnt have to be merged back in like hypergrid for the avination code
[11:52] Gavin.Hird the Kokua OS codebase is now about 2300 commits different from FS for instance because we have not taken a lot of the SL code
[11:52] steve Franklin: My thought on that is that if extra stuff slows down the viewer it should be removed. Happiness is tight code!
[11:52] Gavin.Hird agreed steve, but if you profile the viewer it spend 89% of the time in the renderer
[11:52] Arielle Popstar: like moses pulling out hypertgrid?
[11:53] Gavin.Hird pathfinding slows it down a tad on region crossings
[11:53] Arielle Popstar: making it useless for most in opensim?
[11:53] steve Franklin: If something is needed later that was in there before it can always be looked at again and possibly be done better
[11:53] Vincent.Sylvester The question Kokua has to ask is how much effort is the codebase worth and what code to keep, because to move forward at magnitude level you will have to throw out of a lot of current code.
[11:53] Arielle Popstar: moses devs said it would take a year to put the code back in Steve
[11:54] Arielle Popstar: a year!\
[11:54] Gavin.Hird the parts I have indentified that clearly can go is Pathfinding and the marketplace code
[11:55] Gavin.Hird those are the biggest sections
[11:55] steve Franklin: moses was designed for a specific purpose
[11:55] Gavin.Hird we have not taken the stupid jelly dolls
[11:55] Vincent.Sylvester Beyond that what worries me more is what if Kokua and OpenSim core starts working together and for some reason Kokua dies out, any changes making SL viewers incompatible with OpenSim will need to be made fully documented because in the event another dev needs to pick it up they can't spend a year trying to learn the code.
[11:55] Gavin.Hird or the nagging trash emptying code as examples
[11:55] Arielle Popstar: yes Vincent
[11:55] Andrew Hellershanks: Melanie was talking about doing an OS specific viewer. I don't remember if she was the one talking about making a modular viewer that would make it easy to add/remove features.
[11:56] Gavin.Hird Melanie has talked about that for years
[11:56] Arielle Popstar: they did do one off of singularity
[11:56] steve Franklin: Good point Vincent
[11:56] Bill.Blight breaking silence : "When SL moves to their new protocols, SL viewers will not work with OS anyway."
[11:56] Gavin.Hird I have cleaned up the existing viewer code and prepared for even more cleanup to get rid of irrelevant sections
[11:56] Vincent.Sylvester It is a unique situation that only a few other opensource projects share, most come with client and server made by the same people, developed alongside one another
[11:57] Gavin.Hird which is why we forked the OS code for Kokua Bill
[11:57] Bill.Blight was my point
[11:57] Gavin.Hird it can connect to SL a few months more, but not very much longer
[11:57] Gavin.Hird the SL version of Kokua cannot even connect to OpenSim any more
[11:58] Gavin.Hird they are literally miles apart already (about 2300 commits)
[11:59] steve Franklin: that is quite a bit
[11:59] Gavin.Hird it is
[12:00] Bill.Blight This just highlights what I said earlier, the viewers need to follow their intended target server, not the other way around ..
[12:00] Gavin.Hird absolutely
[12:00] Vincent.Sylvester An idea to solve that would be not making Kokua the only candidate for an OS-only version. I am personally pretty attached to Firestorm and frankly feel that if OpenSim core really goes the direction of giving SL compatibility the finger then they should all be given the information and potential help to make their viewers compatible. Not just for fairness, but because it increases redundancy should one of them decide its no longer worth building for OpenSim.
[12:01] Bill.Blight FS has already said they are sticking to SL
[12:01] Gavin.Hird the resource requirement to keep everything in one viewer is getting very high
[12:02] James atLLOUD: Does indicating Kokua as a reference viewer make it exclusive?
[12:02] Vincent.Sylvester Firestorm says a lot when the days get longer I would not rule anything out just straight of the bet
[12:02] Andrew Hellershanks: If SL and OS diverge in their protocols you would be splitting the viewer dev teams in two and both the SL and OS versions of a viewer would likely suffer as a result.
[12:02] Gavin.Hird The KokuaOS code base can be forked by anyone, but it would be more helpful if more people worked on it
[12:02] Bill.Blight Agreed Andrew
[12:03] James atLLOUD: I thought Firestorm was going to drop OS months ago due to lack of OS devs. Still works for me tho.
[12:03] Gavin.Hird to make the macOS version is almost a full time job for me
[12:03] Freaky.Tech Andrew Andrew, you did not notice that viewers diverged at the time when capabilities were introduced
[12:03] Gavin.Hird it is impossible for one person to maintain a macOS, Linux 32, Linux 64, Win 32 and Win 64 bit version
[12:03] OtakuMegane Desu: How many devs for the other viewers are actually handling OS stuff though?
[12:03] Bill.Blight James, that is because SL had not switched to the new protocols yet, when that happens the FS versions for SL will not work with OS
[12:03] Arielle Popstar: one for FS i thought
[12:04] Vincent.Sylvester All I am saying is that any viewer should be given the info they may need to keep compatible and the focus should be on inclusion
[12:04] Gavin.Hird FS has up to 7 developers
[12:04] Gavin.Hird they also work very close with LL
[12:04] Bill.Blight I should say FUTURE FS SL versions
[12:04] Arielle Popstar: agreed Vincent
[12:04] Gavin.Hird they have access to Havok
[12:04] steve Franklin: Vincent has a good point
[12:04] Gavin.Hird for the SL builds, we don't
[12:04] Gavin.Hird they have a KDU license, we don't
[12:05] Bill.Blight Ummm viewer devs can go look at the code, if they are a dev, they will see what they need, I am sure Gavin has done this .. Not like the OS source code is hidden ..
[12:05] James atLLOUD: ty @bill
[12:05] Gavin.Hird there is now an agreement that LL can take any FS patches and work into the SL viewer
[12:05] Vincent.Sylvester Any divergence should be documented, the viewers should be given notice of changes and assistance in understanding the changes made equally. If they decide not to care that is on them, but the more the merrier.
[12:06] Vincent.Sylvester Approaching people with open arms can go a long way to change someones perception :)
[12:06] Leighton.Marjoram agree Vincent, no needs to burn bridges not a good ground for collaborative working relationships
[12:07] Gavin.Hird most viewer development is UI modifications
[12:07] Gavin.Hird very little is done to the internal workings of them
[12:07] Gavin.Hird flüstert: all that code comes from LL
[12:07] Andrew Hellershanks: Bill, viewer devs and OS devs can look at each others code but it takes time to figure out another code base. A person would also need to have some familiarity with both C# and C++.
[12:08] Gavin.Hird The KokuaOS code is now different not so much because of what we added, but what we stopped adding
[12:08] Bill.Blight Well yes Andrew that is kind of an assumption if you are going to take on a project of that scale ..
[12:08] Gavin.Hird and Objective-C
[12:09] Gavin.Hird and python
[12:09] Gavin.Hird and Cmake
[12:10] Andrew Hellershanks: Bill, partly a question of scale. If you have the two projects working with each other (via docs?) then the people in one camp don't need to spend time trying to make sense out of code in the other camp.
[12:10] Vincent.Sylvester While it is true that most viewers take LL code and put their mods ontop there are a few who have actually implemented some new features, nothing big, certainly nothing the size of a complete protocol change, but enough to require a decent amount of programming knowledge so I would put it past some of them to be capable of dealing with new OpenSim protocols.
[12:10] Gavin.Hird and not the least be able to test on Windows 7, 8, 10 10 S
[12:11] Gavin.Hird macOS 10.11 10.12 10.13
[12:11] Bill.Blight You don't just pickup the source one day and say, "I'm going to make this better because I know better" , unless you actually know what you are doing, or planning on learning.
[12:11] Gavin.Hird a myriad of Linux configs both 32 and 64 bit
[12:13] Gavin.Hird a viewer build has about 15000 files involved
[12:13] Gavin.Hird that excludes the prebuilt libraries
[12:13] Vincent.Sylvester You say that like its a limiting factor
[12:14] Bill.Blight it is limiting based on the skill set of the person looking at it
[12:14] Gavin.Hird it is a limiting factor in that it takes a lot of time to understand what's going on
[12:14] Gavin.Hird even more to make fundamental changes
[12:15] Gavin.Hird which is why any TPV only scratch the surface
[12:15] steve Franklin: yes makes sense
[12:16] James atLLOUD: Opensim only has TPV, no?
[12:16] Gavin.Hird there are no pureplay opensim viewer
[12:16] Vincent.Sylvester If that is such a problem then maybe the aim should be to start from scratch and only pull in the parts needed. Yes that is also a lot of work, but at the moment there is no real alternative other than putting in a lot of work to clean it all up.
[12:17] Vincent.Sylvester All the more reason to work together and attempt to get the whole thing more streamlined.
[12:17] Gavin.Hird virtually all parts are needed
[12:17] Gavin.Hird the most tricky part is the renderer
[12:17] Gavin.Hird which is a very highly optimized piece of code for rendering SL /OpenSim primitives
[12:17] OtakuMegane Desu: Stating from scratch has been proposed for a very long time. Very often unless a codebase is truly abysmal it often is still simpler to refine and clean up than start anew.
[12:18] Gavin.Hird even LL has not had the resources to rewrite it
[12:18] Gavin.Hird so right now we are stuck with very core parts of the LL code if we want any viewer at all
[12:19] Arielle Popstar: radegast
[12:19] Arielle Popstar: put a good renderer into it
[12:19] Gavin.Hird radegast does not render anything
[12:19] Arielle Popstar: then all is C#
[12:19] Gavin.Hird you don't write a renderer in C#
[12:20] Gavin.Hird it is not even possible
[12:20] Arielle Popstar: ok
[12:20] Selby.Evans Very useful discussion -- sorry I must go -bye all.
[12:20] Arielle Popstar: but there is a rudimentary one in it
[12:20] Gavin.Hird you described it correctly
[12:20] Andrew Hellershanks: it is getting to that time of day. Any other OS topics for today before we start losing a lot of people to RL?
[12:20] Gavin.Hird do you want opensim to be rendered rudimentary?
[12:21] Arielle Popstar: well no
[12:21] Gavin.Hird there you have it
[12:22] Arielle Popstar: just throwing out possibilities
[12:24] Andrew Hellershanks: At least we were not talking about physics for a change. :)
[12:25] James atLLOUD: lol -yeah, no meshies chat.
[12:25] Arielle Popstar: hehe
[12:27] Vincent.Sylvester How should I put this, a renderer rewrite is a lot of work, but the least that could be done is explore and collect information on which available engine out there would be the most promising to attempt a rewrite in, thats my opinion anyways.
[12:28] Vincent.Sylvester It needs to happen eventually anyways and while its low priority now discounting it completely just makes it harder later on.
[12:28] Andrew Hellershanks: It would also depend on how many hooks there are between renderer and rest of the viewer code.
[12:28] Arielle Popstar: good points Vincent]
[12:29] Vincent.Sylvester At the moment the most that can and should be done is improve the "backend" aka network handling, data storage and such systems that make the viewer "act" faster
[12:29] Andrew Hellershanks: Gavin, I suspect it might be a bit like hooking in a new physics engine.
[12:29] Gavin.Hird you can speed it up a little bit, but about 89% of the CPU time is spent in the renderer
[12:29] OtakuMegane Desu: While an improved renderer would be lovely, it's a huge undertaking. There's a reason most games build on an existing system rather than roll their own.
[12:30] Gavin.Hird LL is working on a change to the cache to use uncompressed data
[12:30] Arielle Popstar: not really and certain parts of neighborhood are still closed off
[12:30] Gavin.Hird they say it laods 3x as many textures per time unit
[12:30] Andrew Hellershanks: I'm not sure what changed in the viewers but it seems the recent versions seems to impose a higher CPU load on a machine than they used to do.
[12:31] Vincent.Sylvester When I work in php I always start two fold, the visible parts on the page and logic underneath. If one of the two sucks at least I have one that works well and compensates for my shitty coding
[12:31] Gavin.Hird they actually do Andrew
[12:31] Arielle Popstar: LL was also looking at changing where it pulls it from. Instead of from the region one is jumping to but instead the one it left
[12:31] Vincent.Sylvester Improve what can be improved until there is no more, then rewrite and pull what can be pulled.
[12:32] Gavin.Hird which is where the incompatibility between the SL and OS version originates from
[12:32] Gavin.Hird OpenSim grids will not be able to use Content delivery networks (CDN) such as Akamai
[12:32] Gavin.Hird unless it becomes sizeable like SL
[12:33] Vincent.Sylvester CDN for which part?
[12:33] Gavin.Hird CDN for all resources loaded by the viewer
[12:33] Gavin.Hird textures, sounds, meshes
[12:33] Vincent.Sylvester You can already reliably use it for assets
[12:33] Gavin.Hird gestures
[12:34] Arielle Popstar: the new way would benefit hypergrid
[12:34] Arielle Popstar: and was what the original os devs wanted
[12:35] Gavin.Hird so which grid can afford to run off-of a CDN?
[12:35] Gavin.Hird flüstert: standalones?
[12:36] Gavin.Hird Akamai has thousands of servers around the planet
[12:37] Arielle Popstar: well in any case it is not through the region one is jumping to
[12:37] Gavin.Hird you distribute your content to them
[12:37] Gavin.Hird and the content will be pulled from the server that is closest and least busy
[12:38] Vincent.Sylvester CDNs are not backbreaking to run, in fact ZetaWorlds has been setup that way for months it's not expensive if you have a decent provider
[12:38] Vincent.Sylvester if you are only looking at akamai then there is your issue, only fortune 500 can afford that stuff
[12:38] Arielle Popstar: so apache type servers?
[12:38] Vincent.Sylvester Plus building your own CDN from a couple VPS servers here and there isn't rocket science either
[12:38] Gavin.Hird if you get big enough, you can do it but for the hobbyist or small provider...
[12:39] Vincent.Sylvester Well thats my personal opinion, but if you want SL for 20 bucks a month that ain't going to happen so spending a bit of money is required, but if done properly that money is easily recovered
[12:40] Vincent.Sylvester If osgrid wouldnt spend a grand on servers each month even they could setup a cdn
[12:40] Arielle Popstar: nod
[12:41] OtakuMegane Desu: They'd just have to spend the money on the CDN instead. It still comes from server(s), just a different setup.
[12:42] Arielle Popstar: be less hassle with hypergid authentications
[12:42] Vincent.Sylvester True, but it would improve performance massively compared to what it is now
[12:42] Vincent.Sylvester takes me a good 20mins for this region to load alone
[12:42] Arielle Popstar: ouch
[12:44] Vincent.Sylvester I actually spent a few hours working out a plan to save osgrid half its operating cost and improve performance
[12:44] Vincent.Sylvester though I doubt they are actually interested in that, I seem to be talking to brick walls thus far
[12:44] Arielle Popstar: simian grid architecture was using apache servers for distribution but core devs rejected going that route because of the difficulty for end users to set up
[12:45] Arielle Popstar: or so they said

Personal tools
About This Wiki