Difference between revisions of "User talk:SignpostMarv Martin/Archive/Implementing new features"

From Second Life Wiki
Jump to navigation Jump to search
(redirect)
 
(15 intermediate revisions by 5 users not shown)
Line 1: Line 1:
== Alternate Rendering Engines ==
#REDIRECT [[:Category:Feature Requests]]
* Heh, isn't SL already running on an open source 3D engine? :-) [[User:Eddy Stryker|Eddy Stryker]] 19:55, 9 January 2007
 
== Better Sound Support ==
* '''Seconded'''. [[Community Bounties#VLC instead of QuickTime]] might help. [[User:SignpostMarv Martin|SignpostMarv Martin]] 07:51, 9 January 2007 (PST)
 
 
== Removing Texture Loading ==
* '''Comment''' How complete do you want this to be? I made a very quick and easy patch that prevents the client from requesting any image downloads, but in my experience it still rendered a few textures like the grass terrain and avatar hair. It sped up the framerate a lot, but didn't change the GPU requirements to run SL. If the latter is your goal you'd probably have better luck by removing shaders and disabling OpenGL extensions. --[[User:Eddy Stryker|Eddy Stryker]] 12:01, 9 January 2007 (PST)
 
* '''Seconded''' -- Disabling texture download/display would be a great asset to people interested in communication more than shopping/etc. [[User:Kamilion Schnook|Kamilion Schnook]]
** As long as you like the color grey :-) --[[User:Eddy Stryker|Eddy Stryker]] 12:01, 9 January 2007 (PST)
* '''Comment:''' As far as I was aware, libSL clients don't load textures, due to the lack of JPEG 2000 support :-P [[User:SignpostMarv Martin|SignpostMarv Martin]] 07:51, 9 January 2007 (PST)
** This is somewhat incorrect. libsecondlife has had JPEG2000 support for quite some time through our jasper shim called libjaspernet, and some projects use it. Most don't however, as there are few interactive GUI clients using libsecondlife. --[[User:Eddy Stryker|Eddy Stryker]] 12:01, 9 January 2007 (PST)
 
 
== Patching it so that llLoadURL opens the F1 Help Browser ==
* '''Seconded''' [[User:SignpostMarv Martin|SignpostMarv Martin]] 07:51, 9 January 2007 (PST)
* Streaming media urls can be implemented in this manner by making use of string concatenation, e.g. <nowiki>llParcelMediaCommandList([PARCEL_MEDIA_COMMAND_URL,'http://' + (string)llGetOwner() + ':' + stringVar__password + '@icecast.example.com:8000',PARCEL_MEDIA_COMMAND_AGENT,llGetOwner()]);</nowiki> [[User:SignpostMarv Martin|SignpostMarv Martin]] 07:51, 9 January 2007 (PST)
* The problem with this, is that doing this via serverside script is that there's only one media URL, and you'd be connecting to it with the owner's key -- What is wanted is each individual client responds with it's own key. For instance, integrating a "Who is listening in SL" to an icecast server. Each connected client supplies it's own key/name via rewriting, and a name2key/key2name DB displays each individually connected user's SL information. [[User:Kamilion Schnook|Kamilion Schnook]] 20:35 Jan 8, 2007 PST
**Using the PARCEL_MEDIA_COMMAND_AGENT, I'm told you can give each agent their own stream to listen to. So you could do the user method, or more likely, append a query string, e.g:<div style="font-size:140%;"><code class="lsl"><nowiki>llParcelMediaCommandList([PARCEL_MEDIA_COMMAND_URL,'http://' + (string)llGetOwner() + ':' + stringVar__password + '@icecast.example.com:8000',PARCEL_MEDIA_COMMAND_AGENT,llGetOwner()]);</nowiki></code></div>
**[[User:SignpostMarv Martin|SignpostMarv Martin]] 03:31, 10 January 2007 (PST)
 
 
== Adding a Avatar Local Stream Channel ==
 
 
== Peer to Peer Voice over IP using above idea, improved ==
 
 
== More efficient local cache ==
* '''Comment:''' Seronis has some good ideas on how this should be implemented. [[User:SignpostMarv Martin|SignpostMarv Martin]] 07:51, 9 January 2007 (PST)
* '''Comment:''' Definitely. One cache per region, to a configurable cap total, would be an immense load off the asset servers, and off of our poor bandwidth. [[User:Buckaroo Mu|Buckaroo Mu]] 11:30, 9 January 2007 (PST)
 
 
== [http://en.wikipedia.org/wiki/Client-To-Client_Protocol CTCP] protocol layered on IM system ==
* Bleh ? [[User:SignpostMarv Martin|SignpostMarv Martin]] 07:51, 9 January 2007 (PST)
 
 
== Command Line Interface Improvements ==
 
 
== More sophisticated IM  features ==
* '''Seconded''' with a big hell yes. In a large group, it's difficult to keep track of a conversation when there is a mass exodus from the session. It's annoying having the system spam the window. [[User:SignpostMarv Martin|SignpostMarv Martin]] 07:51, 9 January 2007 (PST)
 
== Shortcut/Link/Alias function in Inventory ==
A pure viewer based solution can "only" be a hack at best. A possible implementation could be:
 
#) Change the viewer, so that when a drag and drop operation is performed on a 'no-copy' item, the error is trapped
#) for each error, register the UUID of the object in a special notecard
#) When presenting the contents of a folder in the viewer, preset a mixture of the real contents and the contents of the note card
#) Hide the special notecard
#) Handle folder drops onto the avatar
#) Handle moving of the special links
#) Handle deletion of the special links
#) Handle deletion of the original 'no-copy' item
 
This will only a partial solution, as an scrips running on the servers will still see the real contents of the folder.
 
:Yea, I realized that this would really have to include a new Inventory Type ID - but do the servers do checking for type IDs? This may be something that can be done client-side with either no server-side change or only one very minor one - the assignment of a Type ID. For instance: Assume the new Inventory Type of Link, which simply stores the UUID of the original object. The simplest thing to do with them regarding copying would be to make them either no-transfer, or "delete on transfer" - they simply go away if you try to transfer them. In every other way, it behaves as a normal inventory item. [[User:Buckaroo Mu|Buckaroo Mu]] 09:15, 11 January 2007 (PST)
 
::This definitely could be fudged by using a series of specially named notecards (for configuration data), allowing the feature to be implemented without requiring modifications to the server
::[[User:SignpostMarv Martin|SignpostMarv Martin]] 10:03, 11 January 2007 (PST)
 
:::It ''could be'' fudged with notecards - but how many hacks do you want in SL? The assignment of an additional inventory type at the server end, as far I as can tell from the viewer source, would require probably two lines of code. We need to work with LL to see if they would support this inclusion, rather than make kludges that work but are far less elegant, and may break scripts. [[User:Buckaroo Mu|Buckaroo Mu]] 16:16, 12 January 2007 (PST)
 
::::You're forgetting something: the servers aren't built for dynamic assets. 'Fudging' it would allow the community to decide on features and interfaces while Linden Lab make the significant changes to the server required for this to work.
::::[[User:SignpostMarv Martin|SignpostMarv Martin]] 21:12, 12 January 2007 (PST)
 
:::::If you ''fudge'' this via notecards, serverside scrips working on inventories will see the inventory as it really is. In effect making life complicated for LSL scripts, because now they have to parse a notecard as well.
:::::[[User:Duffy Langdon|Duffy Langdon]]
 
== DirectX3D Hardware Acceleration ==
 
The OpenGL code is scattered all over the entire codebase. You'd have to abstract out all the 3D code first, it wouldn't be an improvement it would be a complete rewrite.
 
== GPGPU Support ==
 
 
== Multiple Monitor Support ==
* '''Seconded''' I've been wanting this since I only had one monitor :-D [[User:SignpostMarv Martin|SignpostMarv Martin]] 11:13, 10 January 2007 (PST)
 
 
== Embedded scripting language for client-side plugins ==
One way of doing this would be to exploit the xpcom architecture of Mozilla. Exposing the viewer core as xpcom interfaces. That way the flexibility and extensibility of the Mozilla engine could be used.
 
The viewer already includes the mozilla engine, using xpcom it will be possible to use Javascript, Java, C++ (Mono/.Net is under way).
 
This would require that a definition of an interface layer between the core viewer and the xpcom system inside Mozilla, once that was defined and implemented, the viewer functionality could be extened using almost any language.
 
[[User:Duffy Langdon|Duffy Langdon]] 12:34, 10 January 2007
 
 
== Nude patch ==
It's inevitable.  Any bets on how long it will be?
# Ages ago. Around about the time alpha textures for clothes were allowed. [[User:SignpostMarv Martin|SignpostMarv Martin]] 00:53, 11 January 2007 (PST)
# Do bear in mind that my mention of a possible implementation for the 'nude patch' would be akin to hallucinating, due to the also mentioned ability to run extra-stringent checks on the server side for submitted content in PG regions. [[User:SignpostMarv Martin|SignpostMarv Martin]] 01:08, 11 January 2007 (PST)
 
I don't think a nude patch could work anyway - the skin and clothing textures are combined on the wearer's client, then the combined texture is uploaded to the server as the single texture that appears on the avatar. That's why "rebaking" your textures on your client affects how other people see you. You could possibly just apply the same skin to everyone but you might just as well leave everyone unrezzed then ;) - Yumi Murakami
: and chase all your payig customers away ? In order for us to build our business out in SL and represent our RL business inside SL such things should never be possible therefor as this is a technical wiki and forum how to further create a better client i suggest to build as much security in the client as possible ro prevent such 'attacks' on our avatars. [[User:River Senyurt|River Senyurt]] 11:34, 12 January 2007
 
:: As I described in the Barriers to implementation section for this "feature", security such as this would have to be built into the server. Since the Viewer is now GPL, all it would take is for someone to remove this 'security' layer, and you're right back where you started, thus Linden Lab are highly unlikely to spend time implementing such a mechanism in the client. The method I described has benefits for both the main and teen grids anyway.
 
:: Aside from the odd oversight over the years (e.g. texture uploads, megaprims), all security and validation operations are executed on the server side. It fits into LL's behaviour to implement such checks on the server, and while they could, and most likely will do something alone these lines, there is unfortunately nothing you can do to stop somebody else making their client get nervous imagine you in your underwear. Or less.
:: [[User:SignpostMarv Martin|SignpostMarv Martin]] 04:18, 12 January 2007 (PST)
 
== Out of sight., out of mind ==
 
# The client has an option to prevent certain types of objects from being rendered
# Objects are addressed server-side by UUID
 
The client would have be made aware of the UUID of an object, categorising it as a new type of object, and a hotkey could be used to toggle the visibility.
 
Similarly, if a photographer or machinimaker is working in-world, the following options would be useful:
 
* Do not render anything other than the land owner's objects
* Do not render objects not owned by people on your friends list
* Do not render avatars on your friends list
 
This would enable people to deal with certain types of griefer attacks more easily, much in the same way that toggling particles on and off can be useful when particle poofers have been dropped.
 
[[User:SignpostMarv Martin|SignpostMarv Martin]] 15:32, 12 January 2007 (PST)
 
=== Visual mute ===
Maybe add Visual Mute, that makes agent's avatar and objects not rendered?
[[User:Tharkang Varayez|Tharkang Varayez]] 04:59, 13 January 2007 (PST)
 
== Night skies ==
 
I'm working on customizing the night sky.  Since the sky, and especially the starfield, are strictly client-side effects, this would involve storing the data in a local file.  Where would be the best place to store this, and what would be the best way to access it?
 
Are there any plans for things like sim-wide data storage?  Such a thing would make it possible to assign per-sim or per-area skies.  I figure a decent starfield would require between 8k and 16k of storage; objects like comets, planets, and the like would be much smaller. --[[User:Carnildo Greenacre|Carnildo Greenacre]] 00:48, 13 January 2007 (PST)

Latest revision as of 07:17, 18 February 2007

Redirect to:

Retrieved from "https://wiki.secondlife.com/w/index.php?title=User_talk:SignpostMarv_Martin/Archive/Implementing_new_features&oldid=10849"