Jump to content

My Browser Builds (Part 2)

roytam1
 Share

Recommended Posts

IXOYE

IXOYE

Posted
Posted (edited)

Hi

I come back to the problem of radio.garden and you will see that not using webgl-angle of google does not satisfy everyone.

On 3/26/2020 at 10:57 PM, grey_rat said:

Hmm :D

Surprisingly, usually the opposite is true OpenGL the problem, not Angle

The decision for http://radio.garden/

GMA4500+, FX5100+ (soft), Radeon 9500+

webgl.force-enabled - true

webgl.disable-angle - true

 

On 3/27/2020 at 1:49 AM, IXOYE said:

Hi

The Grey_rat solution worked.
The problem was OpenGL, I have an Ati radeon HD 4650 AGP graphics card on the pc. The problem is solved on new moon 28.9, Basilisk 52.9 and iceap-upx
"Radio.garden" now works perfectly.
Grey_rat:worship:

I followed the advice of Grey_rat above and it worked perfectly until the day when a tooltip generated by the radio-garden website forbids me definitively to use webgl of my ATI radéon card.

And this, on all Roytam1 browsers except basilisk 55 which works with webgl angle. The consequence is a return to the beginning of my problem.:wacko:

 

The reactivity of radio.garden to modify the use of its website is surprising...
Thanks Google

 

I think you will all understand GOOGLE  webgl.angle is essential, and probably also has spyware for all browsers.:no:

 

update:radio.garden works again under "webgl angle of google" on new moon 28.9, basilisk 52 and iceape-uxp and seamonkey 2.49.5. The responsiveness is incredible. But the webgl ati radeon has been blocked

 

Edited by IXOYE
update

UCyborg

UCyborg

Posted
Posted (edited)

Regarding the problem with Radio Garden, with ANGLE disabled, it bombs out in this part of the browser code: 

#ifdef XP_WIN
        if (gl->GetContextType() == gl::GLContextType::WGL &&
            !gl::sWGLLib.HasDXInterop2())
        {
            DestroyResourcesAndContext();
            MOZ_ASSERT(!gl);

            const nsLiteralCString text("Caveat: WGL without DXGLInterop2.");
            ThrowEvent_WebGLContextCreationError(text);
            return NS_ERROR_FAILURE;
        }
#endif

https://github.com/roytam1/UXP/blob/1ae4416c293f3bdde3d377643692e3d407747142/dom/canvas/WebGLContext.cpp#L966

It's checking for presence of NV_DX_interop2 extension, which isn't supported by any Windows XP compatible graphics driver since the extension is used for sharing Direct3D 10/11 resources with WebGL context.

Omitting the check for this particular extension from the code should make it work (maybe the correct way to go is doing HasDXInterop if we're accelerating the browser with D3D9 and HasDXInterop2 for D3D11).

Basilisk 55 doesn't seem to have that check anywhere, but it's present in current Firefox builds.

Edited by UCyborg
1
Jody Thornton

Jody Thornton

Posted
Posted

Hey guys just a question.  As it stands now in April 2020, is there a still a major compatibilitiy reason to run Serpent 55 as opposed to 52?  I run Windows 8 with the original Basilisk as my backup browser, because I need to use a web application that requires WebRTC?  If I were to use Serpent 55 instead, would there be better compatibility provided?  Is performance better?

Any downsides to Serpent 55 as it stands?

 

Omntech

Omntech

Posted
Posted
10 hours ago, Jody Thornton said:

Hey guys just a question.  As it stands now in April 2020, is there a still a major compatibilitiy reason to run Serpent 55 as opposed to 52?  I run Windows 8 with the original Basilisk as my backup browser, because I need to use a web application that requires WebRTC?  If I were to use Serpent 55 instead, would there be better compatibility provided?  Is performance better?

Any downsides to Serpent 55 as it stands?

 

Pretty much the only one who can answer that question is you, since you need to use a specific web application that requires WebRTC. Run it and see how it behaves under Serpent 55.

 

 

grey_rat

grey_rat

Posted
Posted

IXOYE, add this filter to the ublock
radio.garden##.Modal_isAlert__1mI6e.Modal_modal__P3_V5

Angle can be turned back on, is already working with Angle. webgl.disable-angle false.  But then stop working WebGL2

1
Jody Thornton

Jody Thornton

Posted
Posted
6 hours ago, Omntech said:

Pretty much the only one who can answer that question is you, since you need to use a specific web application that requires WebRTC. Run it and see how it behaves under Serpent 55.

 

 

However, there must be a collective experience thus far as o how Serpent 55 and 52 compare.  And yes, I understand that the whole two threads can be read through, but I'm sure someone can offer up a synopsis.  I appreciate it.

:)

 

IXOYE

IXOYE

Posted
Posted

Hi

3 hours ago, grey_rat said:

IXOYE, add this filter to the ublock
radio.garden##.Modal_isAlert__1mI6e.Modal_modal__P3_V5

Angle can be turned back on, is already working with Angle. webgl.disable-angle false.  But then stop working WebGL2

I disabled webgl2 on newmoon 28.9 with the filter on ublock. It now works without tooltips with webgl ati radeon, but I have slowdowns in music throughput with sudden cuts in radio sound that I can't explain. The CPU load is however the same as on Angle. With GOOGLE webgl angle, I don't have this random slowdown problem. Strange No?

thanks Grey_rat:rolleyes:

Jody Thornton

Jody Thornton

Posted
Posted
On 4/13/2020 at 4:32 PM, Jody Thornton said:

Hey guys just a question.  As it stands now in April 2020, is there a still a major compatibilitiy reason to run Serpent 55 as opposed to 52?  I run Windows 8 with the original Basilisk as my backup browser, because I need to use a web application that requires WebRTC?  If I were to use Serpent 55 instead, would there be better compatibility provided?  Is performance better?

Any downsides to Serpent 55 as it stands?

 

Well I tried it myself.  It tends to record and playback somewhat choppy with/without Direct2D and Hardware Acceleration enabled, so now I know.  Basilisk 52 (as in the original product) it is.

:)

 

roytam1

roytam1

Posted
  • Author
Posted
9 hours ago, fumarpaco said:

@roytam1

Firstly, I've been following your work since the days of hk.geocities.com/roytam1. Great to see you're still around and thanks for all your efforts!

Secondly, I was wondering whether you'd be able to make a build of Pale Moon 3.6.32 with support for modern encryption. This particular browser runs on Windows XP and non-SSE2 processors, has some HTML5 features and security fixes from newer versions of Firefox backported in, and is much faster and lighter than all the alternatives I've tried. I use it for sites where the non-availability of contemporary browser features is not a problem, and results have been great so far. However, I expect lack of updated TLS support to become a problem in the near future, now that old protocols and ciphers are being "retired" from the Web.

I was pleased to find that copying all NSS-related DLLs from your rzbrowser-tls12-20200127 results in TLS 1.2 support! But it does not enable modern algorithms (GCM, ChaCha20, Poly1305, SHA384), and adding the corresponding values to about:config has no effect. Furthermore, every single cipher suite that is supported by this browser is considered either weak or insecure by current standards.

The base source code can be found at https://archive.org/download/Pale_Moon_Archive_2015_08/Pale_Moon_Archive_2015_08.zip/Pale_Moon_Archive%2FSource code%2Fpalemoon-3.6.32-source.7z. Even a one-off release would be enough to make this usable for many years to come :)

Thanks in advance for any answer!

you may just use my 3.6 NT4 fork instead.

1
roytam1

roytam1

Posted
  • Author
Posted (edited)

New build of Serpent/UXP for XP!

Test binary:
Win32 https://o.rths.ml/basilisk/basilisk52-g4.5.win32-git-20200418-714023a-uxp-222d316c8-xpmod.7z
Win64 https://o.rths.ml/basilisk/basilisk52-g4.5.win64-git-20200418-714023a-uxp-222d316c8-xpmod.7z

source code that is comparable to my current working tree is available here: https://github.com/roytam1/UXP/commits/custom

IA32 Win32 https://o.rths.ml/basilisk/basilisk52-g4.5.win32-git-20200418-714023a-uxp-222d316c8-xpmod-ia32.7z

source code that is comparable to my current working tree is available here: https://github.com/roytam1/UXP/commits/ia32

NM28XP build:
Win32 https://o.rths.ml/palemoon/palemoon-28.9.1a1.win32-git-20200418-f58079376-uxp-222d316c8-xpmod.7z
Win64 https://o.rths.ml/palemoon/palemoon-28.9.1a1.win64-git-20200418-f58079376-uxp-222d316c8-xpmod.7z

Official UXP changes since my last build:
- Issue #1512 - Improve handling of multiple selections. (a2bc0e0ec)
- Create SECURITY.md (9f4b2a0e2)
- Issue #1513 - Bail on orphan node (e4548d6ad)
- Issue #1514 - Perform validity check before trying to use the content. (6245e5a8f)
- Merge branch 'master' of https://github.com/MoonchildProductions/UXP (f1675650a)
- Issue #1516 - Part 1: Add a global build define to determine if the client can draw in the titlebar and update consumers of the local define (6f6d1f6ff)
- Bug 1305458 - Changing -moz-appearence on hover breaks change event (5352b69a9)
- Bug 1321284 - Crash in nsCSSFrameConstructor::GetInsertionPrevSibling when trying to reframe native anonymous content (bebec8fcb)
- Issue #1375 - Fix unified sources (b356448d1)
- Bug 1334247 - Remove nsIAnonymousContentCreator::CreateFrameFor (e215bfbb8)
- Bug 1334358 - Remove nsCSSFrameConstructor::CreateAnonymousFrames (4e2b4b9bc)
- Bug 1336011 - Fix Crash in InvalidArrayIndex_CRASH in mozilla::EditorBase::DeleteSelectionImpl (c66ed1275)
- Issue #1517 - Enable high-resolution DOM timestamps on trunk (963fc8079)
- Bug 1331322 - Allow tagging of pseudo-implementing native anonymous content with the pseudo type at creation time (4375774c9)
- Bug 1340333 - Eagerly compute whether a frame is really a browser. (0d149a4c0)
- Bug 1340885 - Null check frame's content when searching for NAC ancestors in GetCorrectedParent (32c3ee92b)
- Issue #251 - Move MOZ_AV1 back to ac configure for everyone to use. (5caf99795)
- Bug 1343937 - Fix a crash in nsWrapperCache.h (1f5c67934)
- Issue #1375 - Overload FromContent() to work with const (2c4d69359)
- Bug 1343879 - Be consistent about the parent style context the document-level anonymous content container should get: it should get no parent style context. (4630e4abb)
- Bug 656197 - Push state updates further out across beforesetattr/aftersetattr (2f59167e6)
- Bug 1347640 - HTMLInputElement shouldn't try to cancel image request on all the type changes (ab05e6f9a)
- Bug 1346623 - Allow anonymous content created with nsIDocument::InsertAnonymousContent can change from non-native to native AC (de45820b6)
- Bug 1330843 - Allow JS to create NAC pseudo-elements (0d362ca50)
- Bug 1296516 - Cleanup a bit of code in layout/base (9f6cb6874)
- Bug 1352389 -Don't push extra script blocker on stack when setting attributes (f07f8aecb)
- Issue #1375 - Include content rules from HTMLTableCellElement::WalkContentStyleRules and move to the mapped attr functionality (5f297c5f5)
- Bug 1355351 - Make pseudo-elements return the correct style via getComputedStyle (7614fdb51)
- Bug 1360157 - Assert that a display: contents child always has a parent (34e2c4054)
- Bug 1361086 - Simplify UndisplayedMap::GetListFor/GetOrCreateListFor to take nsIContent* rather than nsIContent** (0f5dcf963)
- Bug 1363481 - Add the old attribute value as a parameter to Element::AfterSetAttr (091d06b43)
- Bug 1348851 - Use new block when better selection isn't found. (ef689a705)
- Bug 1367683 - Optimize initializing nsRange (940d191ef)
- Bug 1365092 - Move side effects of SetAttr and ParseAttribute to BeforeSetAttr and AfterSetAttr (17f7e1c8c)
- Bug 1337698 - Use UniquePtr instead of nsAutoPtr in editor (516fd67d5)
- Bug 1371170 - Add non-virtual EditorBase::GetSelectionController (5a379a4b1)
- Bug 1372829 - Part 1: Make mozilla::PlaceholderTransaction inherit mozilla::SupportsWeakPtr instead of nsSupportsWeakReference (ea3a2ce27)
- Bug 1332353 - Make it clearer when a stylesheet is really owned by its mDocument (2e2190a50)
- Bug 1372829 - Part 2: mozilla::EditorBase should cache raw pointer of nsISelectionController and nsIDocument with nsWeakPtr (32e815512)
- Bug 1373798 - Move HTML dir attribute state into event state flags (4d1d777e7)
- Bug 1368802 - nsFrameIterator::GetPlaceholderFrame should only try to get the placeholder for out-of-flow frames, because in-flow frames never have a placeholder (2bcd8923f)
- Bug 1368547 - Remove nsFrameManagerBase::mPlaceholderMap and instead store the placeholder on a frame property on the out-of-flow (0c99ad16f)
- Bug 1377648 - Fix HTMLSummaryElement::IsSummary() on removing the element (53428ad3f)
- Bug 1377978 - Make nsRange use uint32_t to offset (70c8ff8e5)
- Bug 1382357 - Wait to destroy frames until after we've successfully fetched the binding (675dce947)
- Bug 1375701 - Atomize class attribute value in the parser in the innerHTML case (f4a1d0123)
- Bug 1389743 - Only reconstruct frames synchronously from ContentRemoved when called from frame construction (e482e335b)
- Bug 1394300 - Streamline nsElementTable.cpp (071957f9f)
- Bug 1396568 - Make DocumentFragment::mHost a strong reference (so that ShadowRoot.host doesn't become null) (8114d58fe)
- Issue #1375 - Stop largely using the parser service (61e46a6d5)
- Bug 1400777 - Slim down nsElementTable.h (a72b82732)
- Bug 1396584 - Remove support for multiple ShadowRoots (5f1294032)
- Bug 1401097 - Simplify gHTMLElements and remove redundant conditions in nsXHTMLContentSerializer::LineBreakAfterClose() (846e0fa4f)
- Bug 1402941 - Add HTMLSlotElement (48f602e65)
- Bug 1381134 - Ensure we're using the correct frame for the :after/:before references (55d83ea6f)
- Bug 1408341 - Implement assignedSlot on Element and Text (375843468)
- Bug 1404789 - Stop reconstructing frames for the whole shadow root each time content is inserted in a shadow tree (ed7faf3fd)
- Bug 1411754 - Rename PresShell::DestroyFramesFor to DestroyFramesForAndRestyle (96dfc63bc)
- Bug 1404842 - Implement Element.attachShadow and Element.slot (873abc7bc)
- Bug 1411878 - Support Element.shadowRoot and Element.assignedSlot / TextNode.assignedSlot on closed shadow root (3a74795a5)
- Bug 1416999 - Remove document.registerElement (5524318fe)
- Bug 1418002 - Remove HTMLContentElement (8beb65dd5)
- Bug 1409975 - Implement node distribution for shadow tree slots (e31ed5b07)
- Bug 1380749 - Retry AdjustAppendParentForAfterContent in case |parentAfterFrame| was a :first-letter frame that we deleted (14bb98e17)
- Bug 979782 - Implement lazy frame construction for display:contents descendants (352fe8398)
- Bug 1413619 - Fix insertion point computation when display: contents pseudos are involved. (ee4857f20)
- Bug 1415152 - No need for AdjustAppendForAfter, since ::after is handled in FindNextSibling (03a898fca)
- Bug 1415538 - Remove dead frame construction code (InsertFirstLineFrames) (e5ca4ada9)
- Bug 1415843 - Remove an unneeded call to AdjustAppendParentForAfterContent (a2cdb9f91)
- Bug 1419964 - Remove AdjustParentFrameForAfterContent (003e537dc)
- Bug 1419799 - Fix nsContentUtils::IsInSameAnonymousTree in Shadow DOM (f0b7ba1e2)
- Bug 1422931 - Fix crash with slot element and make webcomponents preference per-doc (fd7a325bc)
- Bug 1419762 - Return the inline continuation of an IB split when appending (38056aa9c)
- Bug 1422197 - Add fast path to get DocGroup in binding code for [CEReactions] (010f37f47)
- Bug 1412775 - Implement Event.composedPath (7e506bd98)
- Bug 1327185 - Refine EventStateManager::FireDragEnterOrExit (d925bfcc6)
- Bug 1425441 - Move relatedTarget to WidgetEvent (f164962a3)
- Bug 1409976 - Add `slotchange` event (3508e79b1)
- Bug 1426536 - Remove nsContentUtils::IsContentInsertionPoint (b93fb5751)
- Bug 1426503 - Remove DestInsertionPoints stuff (53df3a74f)
- Bug 1417829 - Remove unresolved pseudoclass (f605c68f1)
- Bug 1348481 - Part 1b: Generalize FindDocStyleSheetInsertionPoint so it doesn't require an array of RefPtrs (0f5b78915)
- Bug 1322661 - Expose (non-XBL) style sheets to devtools (f6221f440)
- Issue #80 - De-unify dom/base (16dba9a30)
- Bug 1425769 - Base class for ShadowRoot and Document to manage style state (9e5e58c0f)
- Bug 1360154 - nsIPlaintextEditor might have to have hasText property for UpdateOverlayTextVisibility (6d76ab9b6)
- Issue #1375 - Fix IsWebComponentsEnabled checks (80c024779)
- Issue #1375 - Merge CustomElements pref with Webcomponents Pref (f35aa3e15)
- Merge pull request #1518 from MoonchildProductions/shadowdom-merge (d5102d6be)
- [docs] Update coding style cf. discussion with Tobin. (38f206cb7)
- Bug 1425685 - Implement HTMLSlotElement.assignedElements(); (222d316c8)

Official Basilisk changes since my last build:
- Remove erroneously kept pre-release update disable pref. (352ee7b)
- Change geolocation pref for official releases to https (fdc28db)
- Update back-end branch pointer (94ae402)
- Switch security storage db to SQL (f81229e)
- Ensure AV1 is enabled in all builds by default. (e518a01)
- Issue MoonchildProductions/UXP#1516 - Change CAN_DRAW_IN_TITLEBAR to MOZ_CAN_DRAW_IN_TITLEBAR except for AppConstants in CUI (4bd0d2e)
- Fix broken preferences advanced pane initialization (164cd31)
- Add app-specific moz.configure example line (8006273)
- Update back-end branch pointer (714023a)

Official Pale-Moon changes since my last build:
- Issue #1772 - Only refresh tabs in restore-on-demand mode. (6a5c0e4d3)
- Update back-end branch pointer (140ba4964)
- Issue #1773 - Improve persona styling on Mac (9ac323697)
- Merge pull request #1775 from Lootyhoof/osx-persona (b0ca7bd7c)
- Issue MoonchildProductions/UXP#1516 - Change CAN_DRAW_IN_TITLEBAR to MOZ_CAN_DRAW_IN_TITLEBAR and set it for Windows only (e94f00530)
- Don't set the drawintitlebar attr in gPrivateBrowsingUI unless we want MOZ_CAN_DRAW_IN_TITLEBAR on macintosh (4fc95785e)
- Add MOZ_CAN_DRAW_IN_TITLEBAR to OS X theme (40f64d482)
- Revert "Re-enable sessionstore cache soft refresh" (62fafab3c)
- Merge branch 'master' of https://github.com/MoonchildProductions/Pale-Moon (427ed7ca3)
- Merge pull request #1778 from Lootyhoof/osx-titlebar (0553f440b)
- Update platform branch pointer (unstable 2020-04-16) (f58079376)

My changes since last build:
- reverted "Switch security storage db to SQL"
- dom/canvas: webgl: disable HasDXInterop2 check (a61d9c1f)

Edited by roytam1
6
roytam1

roytam1

Posted
  • Author
Posted

New build of BOC/UXP for XP!

Test binary:
MailNews Win32 https://o.rths.ml/boc-uxp/mailnews.win32-20200418-535efd70-uxp-222d316c8-xpmod.7z
Browser-only Suite Win32 https://o.rths.ml/boc-uxp/bnavigator.win32-20200418-535efd70-uxp-222d316c8-xpmod.7z

source patch (excluding UXP): https://o.rths.ml/boc-uxp/boc-uxp-src-xpmod-20191123.7z

Official repo changes since my last build:
- Resolve un-pp false positive in notification.properties (77170b3c)
- Update gitignore (84ffb496)
- Update platform commit pointer (07064ab6)
- Use nss store 9 (c75a63dc)
- Change some stuff in mach stub (1030b207)
- Update platform commit pointer (cda3df0c)
- Update platform commit pointer (86f7e2f5)
- Remove conditional titlebar drawing from Interlink. (3d43e19e)
- Merge branch 'TRUNK' of https://github.com/binaryoutcast/binoc-central into TRUNK (756e2441)
- Merge pull request #99 from athenian200/conditional-titlebar-drawing-removal (08a39b8a)
- Update platform commit pointer (0bb52dc3)
- Merge branch 'TRUNK' of https://github.com/binaryoutcast/binoc-central into TRUNK (8634803f)
- Update platform commit pointer (535efd70)

My changes since last build:
- reverted "Use nss store 9"

--

New build of IceApe-UXP for XP!

Test binary:
https://o.rths.ml/gpc/files1.rt/iceape.win32-20200418-id-088353e-ia-041bf6a-uxp-222d316c8-xpmod.7z

4
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...