Jump to content

My Browser Builds (Part 3)

roytam1
 Share

Recommended Posts

roytam1

roytam1

Posted
  • Author
Posted
1 hour ago, Slavich said:

Hello, Roytam.

There was a problem with the display of the site https://gosuslugi.ru . This is the largest state portal for servicing citizens of Russia.

For some time now, all Firefox-based browsers cannot display the user's personal account. Screenshot: https://ibb.co/h2t9FWZ

Is it possible to fix this in New Moon?

Thank you for your work. I wish you good luck in all your endeavors.

see https://repo.palemoon.org/MoonchildProductions/UXP/issues/1691

1
Link to comment
Share on other sites

anton12

anton12

Posted
Posted (edited)

Re: Serpent 52 (2022-10-01)
      New Moon 28 (2022-10-01)

Website:

https://www.semanticscholar.org/paper/A-Concise-Review-on-Epigenetic-Regulation%3A-Insight-Golbabapour-Abdulla/7d5284d53170ce6a7ed38798aeb0f8d159365baf


Question:
Who can  open the website above AND click away the ACCEPT & CONTINUE button to gain  FULL access to this website with  the latest versions of NM28 (2022-19-01)  or  St52 (2022-19-01)  ?

Edited by anton12
Link to comment
Share on other sites
roytam1

roytam1

Posted
  • Author
Posted

Off-topic: Thanks for consumption voucher from HK government, and a special offer of HKD 2020 of AMD 5800x, I got a almost free upgrade from 3700x. So there may be about 20% faster for compiling browser's source code.

6
Link to comment
Share on other sites
mina7601

mina7601

Posted
Posted
On 10/1/2022 at 3:45 PM, anton12 said:

Re: Serpent 52 (2022-10-01)
      New Moon 28 (2022-10-01)

Website:

https://www.semanticscholar.org/paper/A-Concise-Review-on-Epigenetic-Regulation%3A-Insight-Golbabapour-Abdulla/7d5284d53170ce6a7ed38798aeb0f8d159365baf


Question:
Who can  open the website above AND click away the ACCEPT & CONTINUE button to gain  FULL access to this website with  the latest versions of NM28 (2022-19-01)  or  St52 (2022-19-01)  ?

Sadly, clicking on the ACCEPT & CONTINUE button does nothing, but the website opens in the latest Serpent version (2022-10-01).

Link to comment
Share on other sites
grey_rat

grey_rat

Posted
Posted (edited)

roytam1, please experiment, change the value in VertexShaderVersion to 000

https://github.com/roytam1/UXP/commit/04671d8d62b1085e8a6e7eaaff4de8efe3ad4ce3

(caps.VertexShaderVersion & 0xffff) < 0x200)

Feodor2 changed the value to 001 for shaders in MyPal 68, but it didn't work for GMA. I noticed that in Firefox 43 software vertex shaders are defined as vs_0_0. Perhaps this is the reason for the problem of acceleration and webgl for GMA

xumd3evotbrsc0qdeg4ro4qm4uw.jpeg

 

Edited by grey_rat
Link to comment
Share on other sites
Multibooter

Multibooter

Posted
Posted (edited)
On 9/24/2022 at 12:00 AM, roytam1 said:

NM28XP build:
Win32 https://o.rthost.win/palemoon/palemoon-28.10.6a1.win32-git-20220924-d849524bd-uxp-f84bc6d60-xpmod.7z
Win32 SSE https://o.rthost.win/palemoon/palemoon-28.10.6a1.win32-git-20220924-d849524bd-uxp-f84bc6d60-xpmod-sse.7z
Win64 https://o.rthost.win/palemoon/palemoon-28.10.6a1.win64-git-20220924-d849524bd-uxp-f84bc6d60-xpmod.7z

 

Hi roytam1,

Your New Moon 28 SSE release of 23Sep2022 WORKS GREAT on my 22-year-old Inspiron 7500 laptop (Pentium 3, 650MHz, 512MB RAM, non-SSE2, WinXP SP3). It worked fine for buying an airline ticket at ryanair.com and paying for it with paypal, a little slow, but it WORKED! The ability to buy an airline ticket at an ad-heavy site is for me the key test of a fully-working browser.

For comparison, it was not possible to buy the above airline ticket with Safari on a 7-year-old iPad Air with the final iOS 12.5.5, the booking process got hung when selecting a seat, even before proceding to the payment part. Updates for the iPad Air ended in September 2019.

I am using New Moon 28 SSE together with User-Agent Switcher 0.1.8.
http://web.archive.org/web/20170401000000*/https://addons.cdn.mozilla.net/user-media/addons/578976/user_agent_switcher-0.1.8-fx.xpi?filehash=sha256%3Abacb5375024202e65b33dbb04d12684fe947abfb16437dfb0c575a55a670dbd1
-> right-click on 1st capture on 21Aug2017 -> Save Link as, to avoid immediate installation

The setting to "Samsung Tizen OS"  helps making some websites more compatible. v0.1.8 is the last version which works/installs with New Moon 28

BTW, New Moon 28 SSE does NOT load properly links/articles at https://www.msn.com/en-us

Added on 5Oct2022: Serpent SSE of 2022-09-29 loads www.msn.com OK

Tor Browser v7.5.6 (based on Firefox 52.9.0), which also works great on the Inspiron 7500 under WinXP and SSE, loads the site OK,
http://web.archive.org/web/20180706013451/https://dist.torproject.org/torbrowser/7.5.6/torbrowser-install-7.5.6_en-US.exe
other versions and languages:
http://web.archive.org/web/20180706013451/https://dist.torproject.org/torbrowser/7.5.6/

I am using NoScript "Classic" (5.1.9) for turning off JavaScript in New Moon 28 SSE.
https://noscript.net/getit/

New Moon 28 SSE release of 23Sep2022 is a must-have for SSE retro-computers. Please keep on updating the SSE version.

T H A N K  Y O U !! for your outstanding work.
 

Posted on the 22-year-old Inspiron 7500 with New Moon 28 SSE

Edited by Multibooter
Added info
Link to comment
Share on other sites
XPRTM

XPRTM

Posted
Posted
20 hours ago, grey_rat said:

roytam1, please experiment, change the value in VertexShaderVersion to 000

 

 

 This wont help. The device creation code is wrong, as it forcefully tries to create a mixed vertex processing device(without checking if Hardware T&L is available). Mixed vertex processing requires hardware vertex processing, and the GMA3000 and older intel iGPUS can't do hardware vertex processing, thus device creation fails.

2
Link to comment
Share on other sites
grey_rat

grey_rat

Posted
Posted (edited)
Quote

Mixed vertex processing requires hardware vertex processing

The browser has D3DCREATE_MIXED_VERTEXPROCESSING

For GMA requires the option to change to D3DCREATE_SOFTWARE_VERTEXPROCESSING ?

Quote

checking if Hardware T&L is available

is it possible to add it to browsers?

Edited by grey_rat
Link to comment
Share on other sites
XPRTM

XPRTM

Posted
Posted (edited)

Yes to both questions. Anyway, just give me some time, i will try to explain better later.

 

EDIT:

here it is. After creating the mD3D9 object, the code(in the DeviceManagerD3D9::Initialize() method) should check if hardware T&L is available, like this: 

    uint32	behaviorFlags = 0;
    D3DCAPS9  pD3D9DeviceCaps;
  
    mD3D9->GetDeviceCaps(D3DADAPTER_DEFAULT,
                        D3DDEVTYPE_HAL,
                        &pD3D9DeviceCaps);
					 
    if (pD3D9DeviceCaps.DevCaps & D3DDEVCAPS_HWTRANSFORMANDLIGHT)
    {
        if (pD3D9DeviceCaps.DevCaps & D3DDEVCAPS_PUREDEVICE)
        {
            behaviorFlags |= D3DCREATE_HARDWARE_VERTEXPROCESSING |
			                D3DCREATE_MULTITHREADED |
                            D3DCREATE_PUREDEVICE | D3DCREATE_FPU_PRESERVE;
        }
        else
        {
            behaviorFlags |= D3DCREATE_HARDWARE_VERTEXPROCESSING | 
			                D3DCREATE_MULTITHREADED |
			                D3DCREATE_FPU_PRESERVE;
        }
    }
    else
    {
        behaviorFlags |= D3DCREATE_SOFTWARE_VERTEXPROCESSING | 
		                D3DCREATE_MULTITHREADED | 
		                D3DCREATE_FPU_PRESERVE;
    }

Then, after checking the underlying hardware caps, simply create a device with the correct parameters, like this: 

   if (!mDevice) {
    hr = mD3D9->CreateDevice(D3DADAPTER_DEFAULT,
                            D3DDEVTYPE_HAL,
                            mFocusWnd,
                            behaviorFlags,
					        &pp,
                            getter_AddRefs(mDevice));

        if (FAILED(hr) || !mDevice) {
          gfxCriticalError() << "[D3D9] Failed to create the device, code: " << hexa(hr);
          return false;
        }
    }

the same thing applies to D3D9ex. Anyway, like i said earlier, the code original code tries to create a mixed vertex processing device,
but it seems like it never switch to software vertex processing mode(via the IDirect3DDevice9::SetSoftwareVertexProcessing method), which is very odd.

Edited by XPRTM
update of content.
1
Link to comment
Share on other sites
grey_rat

grey_rat

Posted
Posted

Firefox prior to version 44 could include hardware WebGL for older GMA. And defines vertex shaders version 0, not 3.

from compositing code (not Webgl) https://github.com/roytam1/UXP/commit/04671d8d62b1085e8a6e7eaaff4de8efe3ad4ce3

(caps.VertexShaderVersion & 0xffff) < 0x200)

200 - Is it still limited shaders to 2.0?

If the browser defines WebGL as VS_0_0 which version it defines for "checking Hardware T&L" and "D3DCREATE_SOFTWARE_VERTEXPROCESSING"?

 

XPRTM, I think you and roytam1 could return the acceleration for GMA :)

 

 

Link to comment
Share on other sites
XPRTM

XPRTM

Posted
Posted (edited)

 Well, fixing it would be rather easy i think, at least for the GM3000. The thing is, i don't have a GitHub account, and i don't want to create one.
I can download the files(only DeviceManagerD3D9.cpp needs to be modified i think), do the proper modifications(with comments and what not), and send it to Roytam via PM, if he's okay with that.  

Quote

Firefox prior to version 44 could include hardware WebGL for older GMA. And defines vertex shaders version 0, not 3.

from compositing code (not Webgl) https://github.com/roytam1/UXP/commit/04671d8d62b1085e8a6e7eaaff4de8efe3ad4ce3

(caps.VertexShaderVersion & 0xffff) < 0x200)

200 - Is it still limited shaders to 2.0?

If the browser defines WebGL as VS_0_0 which version it defines for "checking Hardware T&L" and "D3DCREATE_SOFTWARE_VERTEXPROCESSING"?

    
    
 Let me explain. There are two ways to check devices capabilities in D3D9. The first one is via the IDirect3D9::GetDeviceCaps method, and this one reports the true capabilities of the video card, IE, for the GMA3000, the VertexShaderVersion cap will always return 0. The second way is via IDirect3DDevice9::GetDeviceCaps, this one only work after successfully creating a device, and the capabilities may vary depending on the device creation flags, IE, in the case of software vertex processing devices(like the Intel GMA3000), the VertexShaderVersion cap will actually return 3.0! So it looks like the ANGLE guys are actually using the wrong method to report the caps information. 

Edited by XPRTM
update of content.
1
Link to comment
Share on other sites
roytam1

roytam1

Posted
  • Author
Posted (edited)
15 hours ago, XPRTM said:

I can download the files(only DeviceManagerD3D9.cpp needs to be modified i think), do the proper modifications(with comments and what not), and send it to Roytam via PM, if he's okay with that.

sure, even posting with pastebin will be OK if you want.

EDIT: got a modified source file from @XPRTM and I compiled new xul.dll on top of basilisk52-g4.8.win32-git-20221001-3219d2d-uxp-de51d3878-xpmod-ia32.7z.

to test, download basilisk52-g4.8.win32-git-20221001-3219d2d-uxp-de51d3878-xpmod-ia32.7z and replace xul.dll from http://o.rthost.win/basilisk/basilisk52-g4.8.win32-git-20221005-xpmod-ia32-d3d9test-xul.7z

Edited by roytam1
1
Link to comment
Share on other sites
XPRTM

XPRTM

Posted
Posted (edited)

I tried to use the Insert image from URL post option, but the forum software rejects the links for some reason.

Anyway, here are 2 pics, showing working layers acceleration on Windows XP Version 2002 SP3, and Windows 7 SP1 x64. Keep in mind that, for the GMA3000, layers acceleration needs to be force enabled, to get around the driver blacklist.

UXP.PNG

UXP3.png

Edited by XPRTM
update of content.
Link to comment
Share on other sites
Guest
This topic is now closed to further replies.
 Share
Go to topic listing

  • Recently Browsing   0 members

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