Jump to content

Firefox 24 - 52 for ME and 98


Recommended Posts

7 hours ago, testaccount66766 said:

Took me a while to get working, but Thanks! It actually works better than the Firefox versions 45 and 48 on the win98 VM, and I'm impressed!

:thumbupThanks for testing and confirmation. Which dll files do you use, of XPSP2 or SP3 or ReactOS or mixed? There are reports about problems with XSP3 files only.

Link to comment
Share on other sites


On 8/19/2021 at 2:37 PM, schwups said:

Common issue: The browser exe causes an error in NVOpenGL.dll after close.

The error no longer wants to occur. As soon as I know exactly how to create it, I will investigate it further.

First Stress Test with KM76. After I pressed the stress button the cube was still spinning. Then after a half or a minute I pressed it again and my machine had frozen.

WebGL Test on P4c(SSE2), P4V800D-X, GF5900XT (OGL1.5 card) system:

Firefox52/Centaury0.17 => Tests fail - no cube

Firefox 9.0.1: https://testdrive-archive.azurewebsites.net/graphics/webglstresstest/:)  => Ok spinning cube

https://get.webgl.org/:}=> inactive cube

Edited by schwups
Link to comment
Share on other sites

6 hours ago, schwups said:

:thumbupThanks for testing and confirmation. Which dll files do you use, of XPSP2 or SP3 or ReactOS or mixed? There are reports about problems with XSP3 files only.

 I used a mixture of the dll files of the XPSP2 and SP3 from ApertureSketch's video:

 

Edited by testaccount66766
Link to comment
Share on other sites

On 8/20/2021 at 8:48 PM, jumper said:

Take a process module Minisnap before exiting and again with the error dialog on screen. Post any diffs and the full error text.

Also try with API logging to DebugWindow.exe and profiled in Depends.

I attached minisnaps before and with and the Faultlog. The error after close occurs, if I run two or more webgl tests at the same time in the browser and close it afterwards.

Minisnap and Faultlog

Link to comment
Share on other sites

On 8/21/2021 at 7:31 AM, Goodmaneuver said:

You could try Blender 2.64 or Make Human 1.02 etcetera to see if there is problems with exit there as well.

As well as what Jumper wanted, see if Firefox is still running with Starter or Process Explorer.

I am not familiar with with Blender or Make Human. I don't know how to run it on ME/KernelEX. I downloaded them, started "2.64a.7z", but some menus are missing, others are not functional.

Firefox runs until I click close on the error message. Afterwards no Explorer or Firefox process remain in this case.

Edited by schwups
Link to comment
Share on other sites

It was just a thought but if you are interested I think MakeHumane will need Python installed. Python from the Python web site, choose 2.7.1 https://www.python.org/downloads/release/python-271/ which is the latest one which will install from the MSI. Choose Windows X86 MSI Installer. Then you copy over the files from another installation like XP x86 do not forget Python2.7.dll which will be in System32 and goes in the ME system folder and replace the one in the MakeHumane folder too. I have copied over 2.7.13 but I think the latest 2.7 should work too which is 2.7.18. Apart from that it will need Win2K KernelEx setting and USP10 and I have a later Glu32 than ME original, it might be OK though. I have Blender on BASE setting and it works with menus working but it is not easily learnt and can give a NVopenGl exit error but does exit properly. I have registry config *\NV*.DLL as BASENOHEAP KernelEx setting. The no displaying the Humane character is the video card, Quadro FX5500, it is the only thing that did not work properly yet worked when I swapped monitors as described. Image upload of blender upper right corner with mouse point over the down arrow and top  menu item 3D view.

Bkender.png

Edited by Goodmaneuver
Link to comment
Share on other sites

On 8/22/2021 at 10:14 AM, Goodmaneuver said:

I have to say  that OpenGL 2.01 did not show up as such after removing all NVIDIA files from the system and installing 77.72, it only showed 1.1. After installing 82.16 I got OpenGl 1.1 still and then when installing back over 82.69 back to 82.16 I got 2.01. The 82.16 is the same size as 81.98 and looks similar but definitely not the same. Now either 81.98 and 82.16 are of OpenGl 2.01 standard or only 82.69 is.

I compared the the files a bit with FileAlyzer and WinMerge/Notepad++ with plugin "Compare". The import funktions with addresses of 81.98 and 82.16 are exactly the same and the export function names match. 82.69 has been revised, but only one function "SetEvent" was removed or forgotten and none added. So it is certainly still the same version. 81.85 => no SetEvent

Edited by schwups
Link to comment
Share on other sites

7 hours ago, jumper said:

Download page appears, but link is invalid.

You need to accept the rather large green rectangle at first time visiting the site then it does not need to be accepted again as it does not appear after that. All writing will be in German I believe. I downloaded it twice using FF52.9 on Vista.

Edited by Goodmaneuver
Link to comment
Share on other sites

81.89 and 82.16 NVopenGL give similar results to 82.69 when just swapped into the system with the rest of the 82.69 driver left in place. See picture on page 20. NVopenGL is loaded in RAM by the diagnosis program. 77.72 NVIDIA did not show by the diagnosis program, it must have showed results for Microsoft OpenGL32 even though NVopenGL was loaded. So 1.1 was only shown as supported with 12% of vs 1.2. Function calls are appropriate for plugin compatibility for example but not solely a determining factor for the functioning of the DLL.

Edited by Goodmaneuver
Link to comment
Share on other sites

I cannot remember any OGL version information by Everest/Aida that deviates from the manufacturer's Nvidia specs information. I usually remove everything (Nvidia) with DriverCleanerPro and then I installed the other version. Currently I haven't installed 77.72 (81.85,81.98) driver on any computer.

Link to comment
Share on other sites

I've Stellarium 0.11.4 on this machine. According to the log the QTGLPaint engine is openGL2. Driver NV82.16.

Log:

Writing log file to: "C:/WINDOWS/Application Data/Stellarium/log.txt"
File search paths:
  0 .  "C:\WINDOWS\Application Data\Stellarium"
  1 .  "."
Attempting to use an existing older config file.
Config file is:  "C:\WINDOWS\Application Data\Stellarium/config.ini"
OpenGL supported version:  "2.0.1"
Qt GL paint engine is:  "OpenGL2"
Cache directory is:  "C:\WINDOWS\Application Data\stellarium\stellarium\cache"
Sky language is  "de"
Application language is  "de"
Loading Solar System data ...
Loaded 38 / 38 planet orbits from "C:\WINDOWS\Application Data\Stellarium/data/ssystem.ini"
Loading star data ...
"Loading "./stars/default/stars_0_0v0_1.cat": 0_0v0_1; 5013"
"Loading "./stars/default/stars_1_0v0_1.cat": 1_0v0_1; 21999"
"Loading "./stars/default/stars_2_0v0_1.cat": 2_0v0_1; 151416"
"Loading "./stars/default/stars_3_1v0_0.cat": 3_1v0_0; 434064"
"Loading "C:\WINDOWS\Application Data\Stellarium/stars/default/stars_4_1v0_0.cat": 4_1v0_0; 1725497"
"Loading "C:\WINDOWS\Application Data\Stellarium/stars/default/stars_5_2v0_0.cat": 5_2v0_0; 7669011"
"Loading "C:\WINDOWS\Application Data\Stellarium/stars/default/stars_6_2v0_0.cat": 6_2v0_0; 26615233"
.
.

Now back to WebGL:

And I did make the stress with Firefox 52. Pressed the button once, the browser and system remained functional. Then I switched to https://webglsamples.org/. I tried first 13 samples of these 23. The result is 5 of 13 failed by browser crash. Aquarium, City, Dynamic Cubemap, Field and Fishtank. The others worked.

Link to comment
Share on other sites

WebGL 1.0 based on OpenGL ES 2.0 and requires OpenGL 2.0 graphics driver support.

My conclusion: It works, but limited, probably due to graphics drivers infirmities, by forcing it, quite likely with Nvidia Geforce 7 (6?) series cards (full support OpenGL 2.0) and drivers NV81.85 and higher only. Browser crashes are possible. It can be assumed that first OpenGL 2.0 drivers do not support the OpenGL ES 2.0 specifications 100%. Therefore I do not give the recommendation to switch it on permanently. Driver ATI Catalyst 6-2 (February 2006) 98/ME supports OpenGL 1.5 only.

Now ANGLE (Almost Native Graphics Layer Engine): WebGL-Renderer    Google Inc. -- ANGLE (NVIDIA GeForce 7800 GT Direct3D9 vs_3_0 ps_3_0)

(Backend Renderer uses Direct3D 9.0c)

It don't rely on OpenGL graphics drivers.

layers.acceleration.force-enabled   true  (Hardware acceleration)

webgl.angle.try-d3d11                  false

webgl.force-enabled                     true   (to ensure ANGLE isn't blocked)

webgl.disable-angle                      false

I ran the first 13 samples in Firefox 52 again. Firefox didn't crash anymore. Two of them ran faulty and Halo gave a blank page. The others seem to be OK. The webgl tests with the spinning cubes failed (no cube) with this setting. So this also isn't perfect, but probably more stable.

Firefox up to version 34: Boolean is webgl.prefer-native-gl and not webgl.disable-angle

P4c(SSE2), P4V800D-X, GF5900XT (OGL1.5 card), driver NV82.69: Firefox 52 - Initialisation of ANGLE failed also with force

I read on FAQ's for Stellarium:  ANGLE needs "hardware providing DirectX VertexShader 2_0 and PixelShader 3.0 (vs_2_0 ps_3_0), i.e. DirectX10. This should include even earlier graphics cards: ATI/AMD X1xxx, NVidia 6xxx, Intel GMA X3000."

My 5900XT has vs 2.0a/ ps 2.0a only.

By the way yesterday I tried to run Stellarium ANGLE version and was able to run 0.13.3-ANGLE-win32 ( machine with GF7800GT). I noticed an obvious flaw that the vertical Menu bar has no buttons and is inoperable.

Edited by schwups
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...