gmane.comp.lang.smalltalk.croquet.devel

Re: Hard to find instance variables in Croquet.

askoh — 2008-12-03T17:44:56

Thanks. When stepping through code with the debugger, do those "fields" behave correctly? How do I inspect their values in the debugger? Aik-Siong Koh John Dougan wrote:

Re: Hard to find instance variables in Croquet.

John Dougan — 2008-12-03T09:41:29

They're almost certainly not instance vars, but "fields". Fields are something like instance vars but they are represented in the myProperties dict in the CObject. If a field is not used it won't take up storage in the dict. Tweak uses a modified compiler that translates variable reference syntax into accessor message sends and creates accessors automatically so that the vars are never actually accessed directly. So in Tweak: (assuming that foo is not a method temporary var) foo := self bar: 4 + 5. self foo: (self bar: 4 + 5). do pretty much the same thing in the bytecode. This is useful as it means that it's hard to accidentally bypass the event mechanisms in Tweak. If you browse a subclass of CObject and right click on one of the instance vars you'll see a pop up giving you the options for the var. The color/emphasis of the var name tells you what kind of var it is. Some description of the fields is at http://tweakproject.org/TECHNOLOGY/Tutorials/BankAccountTutorial/ and at http://tweakprojec

Hard to find instance variables in Croquet.

askoh — 2008-12-03T07:46:29

There seem to be many instance variables in the class CPlayer that are not showing up in the debugger and 'show hierarchy'. Why? Is this a known bug or something special? Aik-Siong Koh

FreeType fonts

Steve Wart — 2008-11-29T00:56:54

I managed to get the FreeType fonts package installed in John's latest cobalt image. It was a bit of a hassle. The SqueakMap gave me a bunch of errors when I loaded the 0.4 FreeType Plus package, and I used the list to copy and paste the offending methods from a Squeak 3.10 image, but it worked. I was actually very impressed that it was able to change fundamental behavior on the various scanner classes without kicking the image out from under me. Kudos to whoever figured out how to make that work! Now I would like to package my changes back up and publish them to a monticello repository somewhere. I thought I'd be able to just reload the packages and find them in an MC change set. Any suggestions on how to do this? I probably loaded this up completely incorrectly, but my fonts look better now. They do seem to be a bit fuzzy but I'm not sure if that's the fonts, my monitor or something I did to my eyes in the past hour :) Steve

New Cobalt Release for 20081120

2008-11-26T15:12:05

Hi, I noticed a little less system load (minus 5-8%) with this new version. My machines specs are... Celeron-M 2GHz, 1024Mb, GeForce FX5200 and I'm running Cobalt on Ubuntu Linux 8.04 with the latest binary drivers from NVidias website. Regards... Th. Braun

OpenAL Harness

Steve Wart — 2008-11-21T18:34:41

Did you ever wonder if OpenALHarness belongs in the CroquetHarness class hierarchy? I wanted to test the sound stuff independently of Croquet, so I made a sound player class and I needed a harness to hold the OpenAL so I copied OpenALHarness out and deleted the voice chat code that was referencing the avatar instance variable. It was easy to do, and I had the impression that it was never designed to fit that way to begin with. It seems the original CroquetHarness is quite nice with lots of helpful comments and it doesn't seem that a deep hierarchy of classes is the right way to organize this code. Has anyone been looking at refactoring this? Should I give it a go? Or maybe it's a paragon of clarity and I just haven't yet achieved enlightenment... Steve

Re: TBoundSphere performance

Phua Khai Fong — 2008-11-17T21:53:33

Since we are on the topic of TBoundSphere, dont mind me asking one question. I have a problem working with TMesh which is related to the TBoundSphere and its picking. My situation is described below:- I have a few meshes created with TMesh by specifying the vertices, normals and texture coordinates. When I add all the meshes under 1 TGroup class, the boundSphere and picking seems to work fine. The problem comes when I have a hierarchy of several TGroup classes and the TMesh are added into the lowest part. An example would be Model>Geo>Mesh>Triangle where Model, Geo and Mesh (there can be several Geo and Mesh objects) are subclasses of TGroups and Triangle is the created TMesh. When done this way, the TMesh objects cannot be picked or displayed properly (invisible at certain angle and distance), I suspect its a problem with the TBoundSphere. Anyone encounter this problem? Additional information to this scenario :- -initBounds is sent to the Model class only after all the TMesh objects are created and

Re: TBoundSphere performance

Steve Wart — 2008-11-17T20:08:08

Thanks David, I am also thinking of pre-calculating the spheres to cache them with my meshes. The hit-testing is fast, it's the initial creation that's expensive. Moving this complex code into C is a bit daunting, although I found a few free(ish) libraries that can take an arbitrary vector of 3d points and return the center and radius of a minimal (or close to it) sphere containing the points. Since we are doing most everything else in Smalltalk that might also provide a nice win. We are creating a large number of objects in this method, but as you say, it is only invoked at certain times. I am activating about 200 frames when I enter a new space and the bounds init is taking on the order of 500ms to 3000ms for each one. Steve On Mon, Nov 17, 2008 at 11:53 AM, David P. Reed reed.com> wrote:

Re: TBoundSphere performance

David P. Reed — 2008-11-17T19:53:09

It's not surprising that so much time is spent on this. It takes the mesh and creates a tight sphere-covering to accelerate ray intersection testing. It's possible to trade space for time by storing the resulting sphere-tree with the mesh source, rather than recalculating it at load time. But the storage space explosion factor is enormous, and it's not clear whether reading in the sphere-tree wouldn't be slow due to file reading (it certainly is slow to send across a slowish network). You could get a constant factor by just making a primitive out of a few of the inner loops, or even much better improvement by using CUDA organized parallelized code (nice research project). Alternatively, and probably much better, change the ray-mesh intersection code (the only client for the sphere-tree) by subclassing the operation by known techniques optimiized for your mesh's properties. OOP is designed for this! Steve Wart wrote:

TBoundSphere performance

Steve Wart — 2008-11-17T17:48:52

Hi, I have some large meshes that are taking a long time to activate. It looks like a lot of the time is spent in a method called initBounds, which in turn is spending a lot of time in TBoundSphere>>calcTree:faces:depth: Have there been any efforts in optimizing this method (e.g. making it a primitive)? Any suggestions on how to approach this? Slang? FFI? Something more clever? Steve

Re: New Cobalt Release for 20081112

pshouse — 2008-11-15T20:52:00

Here's a quick hack to get him facing the other way: In 'TLoadOgreMesh>>buildMesh' comment out the line 'theMesh addRotationAroundY: 180.0'. It's the second to last statement in the method. Mark P. McCahill-2 wrote:

Strange problem in my harness

Steve Wart — 2008-11-15T01:08:43

I'm trying to get up to speed on the master/participant model in croquet and have run into a few problems with process interactions and was wondering if anyone can suggest what might be going on. I have a method where my AvatarUser is entering a new space and it calls a TTeleportUtilities method makePortalToAnotherWorld. In that method, it tries to connect to a postcard using the following snippet: harness := Processor activeIsland project ownerMorph harness. It quite often comes back with nil. When I step into the code occasionally there's my harness, and occasionally it's not. Any pointers on where I can find out how this stuff is supposed to be wired together? Thanks, Steve

New KMZ Importer Release 0.2

Phua Khai Fong — 2008-11-14T14:59:23

Download the package http://croquet-src-01.oit.duke.edu:8886/KMZImporter/Collada-kayef.22.mcz here Package requires http://croquet-src-01.oit.duke.edu/cobalt.html Cobalt20081112 and above. Demo video can be viewed http://www.youtube.com/watch?v=mGCLMkYYY6s here More info at http://www.oopgames.com/freecad/release-kmz-importer-ver-02 http://www.oopgames.com/freecad/release-kmz-importer-ver-02 Feedback will be most appreciated!! Regards, Fong

Re: FloatingPointException

Les — 2008-11-14T14:57:50

Of course you are right, Josh, I didn't realize that the function in question was a library function, so that is why I made the suggestion I did. Regards, Les H On Wed, 2008-11-12 at 01:05 -0800, Joshua Gargus wrote:

Re: New Cobalt Release for 20081112

John Dougan — 2008-11-14T03:11:12

Rats, I was worried about that. Which hardware do you have? Use glview if you can (link in the 2008-12-11 notice) and post the xml save file, or glxinfo if you can't and post it's output. Thanks, -- John On Thu, Nov 13, 2008 at 19:01, askoh askoh.com> wrote:

Re: New Cobalt Release for 20081112

askoh — 2008-11-14T03:01:58

I got the following error: GL_ARB_vertex_buffer_object is not present

Re: New Cobalt Release for 20081112

Mark P. McCahill — 2008-11-13T16:30:12

On Nov 13, 2008, at 1:19 AM, Andreas Raab wrote: I assume this is because virtual Ninja warriors never willingly turn their backs on players like Andreas who are holding weaponized mice ;-) I think that Ninja is in a total-player-distrust coordinate system - he always faces me, too

Question about tutorial

Parodi Elisabetta — 2008-11-13T11:22:56

Dear Croquet-Dev, I've seen the tutorial at: http://opencroquet.org/index.php/Island_Creator/Blender_Basic_Lessons_TO C it's very interesting and detailed, it's great! The part to which I'm nost interested is the lesson: Blender Basic Lessons TOC (http://opencroquet.org/index.php/Island_Creator/Blender_Basic_Lessons_T OC) Unfortunately it is not complete.. Could you please help me to understand: - how it is possible to create the 3D "panels" containing informations - what else 3D panels can do - which is the effort required to develop custom 3D objects with "custom" behaviour, for example: A panel shows a picture, when the user clicks on it it shows "information" (doc/pdf,etc...) - or if this is already available - if "putting a new object" into the world requires programming, or it could be done via a graphical interface by a not-programmer user. Thanks in Advance, Best Regards, Elisabetta

Re: New Cobalt Release for 20081112

Andreas Raab — 2008-11-13T06:19:04

Much better! And you even got the download link right ;-) The avatar animation works great btw on my little box with no shader support. The only thing I was wondering about is that the default ninja (?) is pointed towards the viewer. I'm not sure if this is intentional or if it's a problem with the software version but other than that this version works fine. Good job! Cheers, - Andreas John Dougan wrote:

Re: FloatingPointException

Joshua Gargus — 2008-11-12T09:05:16

We seem to be veering off into wild speculation. Let's see if we can avoid doing so... There are two methods that Les might be talking about when he was "just examining the code again": - FloatArray>>dot: - FloatArrayPlugin>>primitiveDotProduct: I'm assuming that we're talking about FloatArray>>dot:, since that's the one that is indexed from 1 to size. It's easy to verify that this is the way that FloatArray is supposed to be used. Try evaluating the following expressions: #(1 2 3) asFloatArray at: 1 "answers 1.0" #(1 2 3) asFloatArray at: 2 "answers 2.0" #(1 2 3) asFloatArray at: 3 "answers 3.0" #(1 2 3) asFloatArray at: 0 "error" #(1 2 3) asFloatArray at: 4 "error" So, if we are executing the Smalltalk code in FloatArray>>dot:, we can't do an out-of-bounds access. However, note that under normal conditions, the Smalltalk code in FloatArray>>dot: will never be executed. Instead, it will execute the primitive specified by FloatArrayPlugin>>primitiveDotProduct:. If you're not familiar with

Re: New Cobalt Release For 20081111 - Fixed Link

Andreas Raab — 2008-11-12T07:31:30

This version doesn't seem to work at all - it blows up right away (SqueakDebug.log attached). Cheers, - Andreas John Dougan wrote: Error: attempt to index non-existent element in an ordered collection 11 November 2008 11:28:30 pm VM: Win32 - a SmalltalkImage Image: Croquet1.0beta [latest update: #2] SecurityManager state: Restricted: false FileAccess: true SocketAccess: true Working Dir C:\Cobalt\cobalt-base-current-build-20081111 Trusted Dir C:\Cobalt\cobalt-base-current-build-20081111\andreas Untrusted Dir C:\My Squeak\andreas OrderedCollection(Object)>>error: Receiver: an OrderedCollection('Icon' 'Choose another name' 'Cancel') Arguments and temporary variables: aString: 'attempt to index non-existent element in an ordered collection' Receiver's instance variables: array: #(nil nil 'Icon' 'Choose another name' 'Cancel' nil nil nil nil nil) firstIndex: 3 lastIndex: 5 OrderedCollection>>errorNoSuchElement Receiver: an OrderedCollection('Icon' 'Choose another name' 'Cancel') Argu