My Browser Builds (Part 4)

[roytam1]

New regular/weekly KM-Goanna release:
https://o.rthost.win/kmeleon/KM76.4.9-Goanna-20230916.7z

Changelog:

Out-of-tree changes:
- import changes from `dev' branch of rmottola/Arctic-Fox:
 - Bug 1240411: P6. Clean up OMX headers. r=ayang (114a1df4be)
 - Bug 1230385, part 1 - Use MOZ_CRASH in ContentChild::AllocP*() methods. r=billm (6f8016bb29)
 - Bug 1230385, part 2 - Use NS_WARNING in unimplemented TabChild methods. r=billm (28a7165773)
 - Bug 1234026 - Pass a --display option to gtk_init in content processes. r=karlt (8e3e17858e)
 - Bug 1218816 - Remove useless semicolons. Found by coccinelle. r=Ehsan (e029c51e58)
 - Bug 1240796 - Implement Uint32x4 <==> Float32x4 conversions. r=sunfish (5a42f571ea)
 - Bug 1240796 - Implement unsigned SIMD compares. r=sunfish (93d4979730)
 - Bug 1240796 - Implement Uint32x4 extractLane in Ion. r=nbp (b5b7c782b6)
 - Bug 1240796 - Add Uint32x4 support to jit-test/lib/simd.js. r=bbouvier (187cbbb1d0)
 - Bug 1245547 - Implement RSimdBox for Uint32x4. r=nbp (0186f0355f)
 - Bug 1244254 - Move SimdTypeToMIRType into the header. r=nbp (5b15375c4e)
 - Bug 1217236 - Block trackers loaded by Flash movies. r=gcp (a1318a33da)
 - Bug 1237402 - Allow certain plugins to be loaded in parent process (r=jimm) (e951737778)
 - Bug 1172304 - Fix to handle short read in Plugin code. r=johns (625dadd61b)
 - Bug 377630 - Preventing filename disclosure, by putting downloaded files in a private directory. r=bz (5aca752a5c)
 - Bug 579517 follow-up: Remove NSPR types that crept in (c7cb9ffc11)
 - Bug 1245724 - Make plugin network requests bypass service worker interception. r=ehsan (a0b7fab3ee)
 - Please enter the commit message for your chang Bug 1244254 - Pass a SimdType to inlineSimd(). r=nbp (6fcacd3c5d)
 - Bug 1244254 - Add IonBuilder::unboxSimd(). r=nbp (d5f0922fd9)
 - Bug 1244254 - Check SIMD arguments in IonBuilder. r=nbp (832d38940f)
 - Bug 1244254 - Replace MaybeSimdUnbox with assertions. r=nbp (9c2ad4a8c5)
 - Bug 1244254 - Add SimdType to MSimdBox and MSimdUnbox. r=nbp (74412371f5)
 - Bug 1238003 - Part 1: Add BooleanPolicy. r=jandem (79c1716cac)
 - Bug 1238003 - Part 2: Use Policy in RegExpMatcher and RegExpTester. r=jandem (a2ef8feec4)
 - Bug 1238003 - Part 3: Add test for Policy in RegExpMatcher and RegExpTester. r=jandem (1653bec783)
 - Bug 1244254 - Simplify MSimd* constructors. r=nbp (0ab2efcb4c)
 - Bug 1244889 - Remove trivial SIMD NewAsmJS factories. r=bbouvier (2f9c41713c)
 - Bug 1244828 - Ensure enough ballast space in CallPolicy::adjustInputs. r=bbouvier (d84dae2175)
 - Bug 1244828 - Ensure enough ballast space in AllDoublePolicy::adjustInputs. r=bbouvier (361092db86)
 - Bug 1245421: Remove dead function CoercesToDouble; r=h4writer (502b9efcee) (fc9fb8b315)
- Revert "pref: disable `javascript.options.ion' for the time being since it can cause crash in random memory location with corrupted stack when browsing (affects archive.org)" (2fd5121803)
- import changes from `dev' branch of rmottola/Arctic-Fox:
 - Remove Compositor::mRenderBounds. (bug 1243071 part 1, r=mattwoodrow) (5f82e99655)
 - Use double precision when computing the backdrop copy rect. (bug 1243071 part 2, r=mattwoodrow) (3db95bac33)
 - Add software mix-blend mode support to the D3D9 compositor. (bug 1243071 part 3, r=bas,mattwoodrow) (4a68678ab7)
 - Bug 1231564 - Don't trigger new composites just to repaint debug overlays. r=BenWa (7dfd13163a)
 - Bug 1249312: UseAcceleratedCanvas on Mac should really be UseAcceleratedSkiaCanvas. r=mchang (0b830d9c6e)
 - Fix stale intermediate surfaces in the D3D9 compositor. (bug 1244821, r=mattwoodrow) (1827598eaa)
 - Bug 1224433 - Part 2: Clamp the invalidation rect to values that fit within nscoord. r=rc (3c08f0531e)
 - Bug 1161978 - Use GetEffectiveVisibleRegion() for Layer invalidation r=matt.woodrow (3209eda22d)
 - Bug 1242720 - Followup: reduce code duplication. r=jfkthame (37ff368177)
 - improve gcc11 and later fix for numeric limits (e7a3e86bb7)
 - Bug 1240589 - Import Chromium's upstream rand_util_win.cc to avoid rand_s() crash. r=jld (9be9342c7b)
 - Bug 1240589 - Cross compilation fixup. (714a88c2a3)
 - Bug 1188211 - Fix more constructors in IPC; r=billm (500382e235)
 - Bug 1200131 - Avoid shadowing members in tuple.h; r=njn (0c3f06c4eb)
 - Bug 1205942 (part 1) - Fix overflows in time_posix.cc. r=jld. (18961ef5dd)
 - Bug 1205942 (part 2) - Fix "always true" warning in child_thread.cc. r=jld. (3c21921822)
 - Bug 1205942 (part 3) - Remove unused OS version detection functions from ipc/chromium/. r=jld. (da54914934)
 - Bug 1205942 (part 4) - Remove GetWinVersion(). r=jld. (345daa48bb)
 - Bug 1141565 - TSan: data race ipc/chromium/src/base/histogram.cc:730 Accumulate. r=gfritzsche. (be428d7396)
 - Bug 1240985 - IPC glib deadlock fix (r=jld) (ab640c7c5b)
 - Bug 1241901 part 4 - Stop using nsAutoPtr for holding primitive arrays. r=froydnj (f8c2de6f8b)
 - Bug 1133073 - Use PR_DuplicateEnvironment to avoid post-fork malloc on all Linux platforms. r=dhylands# Please enter the commit message for your changes. Lines starting (58b793b6db)
 - Bug 1215151 - Add XPCOM leak checking for base::Thread. r=jld (5f58074c97)
 - bug 1249622 - make ThreadData::list_lock_ a StaticMutex r=froydnj (1fbc823209)
 - Bug 1213320 - Detect IPC::Channel leaks with the XPCOM leak checker. r=jld (d483f4abc2)
 - bug 1161166 - Move some implementation out of ipc headres to not include headres for EndianU32_* r=billm (6125f1789d)
 - Bug 1221056 - Make the destroy messages for PTexture and PCompositable part of the current transaction when applicable. r=sotaro (52e3533e8d)
 - Bug 1236167: P1. Do not attempt to destroy a non allocated surface. r=nical (169570d50e)
 - Bug 1236167 - Don't attempt to take a compositor snapshot if parameters are incorrect. r=jrmuizel (7b65e2df32)
 - Bug 1229780 - Remove TextureClientPool helper functions from ClientLayerManager, r=mwoodrow (3f8cc506d9)
 - remove code without reference in FF or TFF (0126b4d3f8) (37721bb087)
- [libwebp] Fix OOB write in BuildHuffmanTable. (efdf1d591d)
- [libwebp] Fix invalid incremental decoding check. (08625c2b16)
- import changes from `dev' branch of rmottola/Arctic-Fox:
 - Bug 1230352 - Update to Oculus SDK 0.8.0,r=vlad (ec5404763a)
 - Bug 1237689 - Enable Oculus hardware latency tester r=daoshengmu (a914fb5c78)
 - Bug 1248757 Use string ::Assign() instead of Adopt() when reading blobs as strings. r=asuth (65e361797f)
 - Bug 1248757 followup - Release blob after assignment in DoGetBlobAsString on CLOSED TREE. (8f5d86a27a)
 - Bug 1240583 - Odin: record and assert whether a function is defined yet (r=bbouvier) (693c0be6fe)
 - Bug 1240583 - Odin: store code-range index instead of entry offset in ModuleGenerator (r=bbouvier) (ac14028824)
 - Bug 1240583 - add thunkWithPatch/patchThunk (r=jandem) (9b53ac4d5b)
 - Bug 1240583 - Odin: fix long jumps/calls on ARM for large modules (r=bbouvier) (3f7bacc9cf)
 - Bug 1240583 - Odin: add MacroAssembler::repatchThunk (r=bbouvier) (1336b1492d)
 - Bug 1037483 adopt microformats-shiv for microformats v2 support, r=tantek (c08afa618d)
 - Bug 1224766 - forward callID to disambiguate multiple gUM requests from same window. r=jesup,smaug (547eafdbaa)
 - Bug 1201393. Remove irrelevant ProcessedMediaStream for nsSpeechTask. r=eitan (f17f8e6821)
 - Bug 1240583: Fix non-unified build for fuzzers; r=luke (cc7ea34899)
 - Bug 1212366 - Part 1. Don't call SetAudioOutputVolume if stream is destroyed. r=roc (af1106491c)
 - Bug 1212366 - Part 2. Don't release SourceMediaStream until StreamListener is called. r=roc (08cf9cf62f)
 - Bug 1220320 - implement the nsSupportsWeakReference. r=baku (0cfd32c83a)
 - Bug 1225347 - Apply audio setting to volume parameter of Speak(). r=eeejay (fc6dbd938c)
 - Bug 1228564 - part 1 : revert the changeset of bug 1190040. r=baku. (921a0f7383)
 - Bug 1195051 - Part 3: Test changes; r=padenot (438a73a408)
 - Bug 1195051 - Part 4: Fix a null pointer crash happening after the destination node gets CCed (90fd50c8ac)
 - Bug 1228564 - part 2 : check audio capturing when the agent is registered/unregistered. r=baku. (14f473625b)
 - Bug 1228564 - Follow-up to fix static analysis build bustage. r=me (295d61a1b6)
 - Bug 1247846 - Odin: switch CallIndirect to wasm binary encoding (r=bbouvier) (5997b4c59b)
 - Bug 1247846 - Odin: refactor ModuleGenerator::finish (r=bbouvier) (a71293c284)
 - Bug 1247846 - Odin: refactor error stub generation (r=bbouvier) (5f9f98b215) (9e41bd2207)
- devtools: fix use of genexprs and legacy comprehensions (b8f9cd342d)
- more devtools js fixes (9c3ef02584)
- import changes from `dev' branch of rmottola/Arctic-Fox:
 - Bug 1247104 - BaldrMonkey: Outline a method to reduce indenetation. r=luke (43aae99318)
 - Bug 1247104 - BaldrMonkey: Enclose most of WasmText.cpp in anonymous namespaces. r=luke (50a93211ef)
 - Bug 1247104 - BaldrMonkey: Parsing, encoding, and decoding for load+store. r=luke (bf87e68074)
 - Bug 1247846 - Baldr: add type declaration to text language (r=sunfish) (4333417275)
 - Bug 1245112 - Part 1: Move MacroAssembler::branchPrivatePtr into generic macro assembler. r=nbp (ba4b42bca0)
 - Bug 1245112 - Part 2: Move MacroAssembler::branchPtr into generic macro assembler. r=nbp (863dbbb121)
 - Bug 1245112 - Part 3: Move MacroAssembler::branch32 into generic macro assembler. r=nbp (e8464ab752)
 - Bug 1245112 - Part 4: Move MacroAssembler::branchTest64 into generic macro assembler. r=nbp (6ac2817f78)
 - Bug 1245112 - Part 5: Move MacroAssembler::branchTestPtr into generic macro assembler. r=nbp (5498a461ec)
 - Bug 1245112 - Part 6: Move MacroAssembler::branchTest32 into generic macro assembler. r=nbp (3f80771dfd)
 - Bug 1245112 - Part 7: Move MacroAssembler::branchFloat into generic macro assembler. r=bbouvier (bfe366f03d)
 - Bug 1245112 - Part 8: Move MacroAssembler::branchTruncateFloat32 into generic macro assembler. r=bbouvier (a2953d2eb1)
 - Bug 1245112 - Part 9: Move MacroAssembler::branchDouble into generic macro assembler. r=nbp (68910e9b54)
 - Bug 1245112 - Part 10: Move MacroAssembler::branchTruncateDouble into generic macro assembler. r=nbp (cb176293fe)
 - Bug 1245112 - Part 11: Move generic MacroAssembler methods into check_macroassembler_style block. r=nbp (faa0180389)
 - Bug 1246957: Make the data section require the memory section; r=luke (9337d637ab)
 - Bug 1247846 - Baldr: add indirect function table and call_indirect (r=sunfish) (a379696e7f)
 - Bug 1247247 - Factor essential wasmEval code into public helper function. r=luke (606562c29b)
 - Bug 1240127 - Fold MTableSwitch with constant int32 operand. r=nbp (350accb1cc)
 - Bug 1246658 part 1 - Refactor MDefinition::constantValue and friends. r=bbouvier (fea967ef6e)
 - Bug 1246658 part 2 - Rewrite MConstant::valueToBoolean to not depend on js::Value. r=luke (8b6cddbd7c) (517c54e11e)

* Notice: the changelog above may not always applicable to XULRunner code which K-Meleon uses.

A goanna3 source tree that has kmeleon adaption patch applied is available here: https://github.com/roytam1/palemoon27/tree/kmeleon76

[Mathwiz]

20 hours ago, nicolaasjan said:

There is a good explanation here:

https://stackdiary.com/critical-vulnerability-in-webp-codec-cve-2023-4863/

BTW, official Pale Moon was already fixed yesterday.

Hmm ... looks like GIMP is likely affected too - I'd better check for an update.

BTW the above link does not open correctly in my browser (St 55). St 52 didn't work for me either. I just get a pageful of Javascript source code in both browsers.

Edited September 16, 2023 by Mathwiz

[nicolaasjan]

1 hour ago, Mathwiz said:

BTW the above link does not open correctly in my browser (St 55). St 52 didn't work for me either. I just get a pageful of Javascript source code in both browsers.

Indeed.

Same happening in official Pale Moon here...

But for those who have no access to a "modern" browser, here is a pdf of the page.

[UCyborg]

You've already forgotten about the New CSS Reset concept?

Edited September 16, 2023 by UCyborg

[nicolaasjan]

2 hours ago, UCyborg said:

You've already forgotten about the New CSS Reset concept?

Missed that entirely.

What kind of a mess is that?

Anyhow, I tested with your userscript on stackdiary.com, but it still displayed wrong here.

Added the line:

// @match         https://stackdiary.com/*

Or is there more that needs to be changed?

 

What did work on that site:

layout.css.is-where-pseudo.enabled ---> false

 

Can Moonchild fix this?

Edited September 16, 2023 by nicolaasjan

[PPeti66x]

@roytam1

Newmoon v27 unexpected behaviour: On DuckDuckGo search page the vertical scroll bar can not moved by mouse (scrolling by wheel or by PgUp/PgDown still works). Bug in NM27 or incompatible code in DuckDuckGo? Observed in versions: current(16.09.2023) and 09.09.2023 (2-3 earlier updates were ignored, so these are unknown).

[AstroSkipper]

On 8/16/2023 at 2:15 AM, AstroSkipper said:

The website https://developer.mozilla.org/en-US/docs/Web/CSS/animation causes 100% utilisation of my CPU on my old computer. Animations of any kind are now killers of old, underpowered computers. I have already deactivated various animations in different ways in my browsers. Now, I have also installed the extension Stylem and installed the following CSS style via copy and save from userstyles.org: https://userstyles.org/styles/180927/disable-css-animations-and-transitions-safely 
This actually stops any CSS animation on the site above and brings my processor back to normal. 

FYI, the CSS userstyle https://userstyles.org/styles/180927/disable-css-animations-and-transitions-safely has unfortunately a negative impact in New Moon 28 on the download window which is not flexible anymore when this CSS style is activated. I use the extension GetEmAll!, which adds two entries to the download window, and the extension Grabit, which also adds a further entry to the download window. With the CSS style disable-css-animations-and-transitions-safely activated, the Save and Cancel buttons sometimes disappear completely or are partially cut off. The whole download window seems to be invariable in size. That's why I only use this script on demand from now on. 

Edited September 16, 2023 by AstroSkipper

[UCyborg]

5 hours ago, nicolaasjan said:

Anyhow, I tested with your userscript on stackdiary.com, but it still displayed wrong here.

Added the line:

// @match         https://stackdiary.com/*

Or is there more that needs to be changed?

Yeah, it would have to be re-written since the directives are located in a specific external stylesheet, that script only works if the CSS code is located in the first style element in the HTML document.

5 hours ago, nicolaasjan said:

Can Moonchild fix this?

They'll have to add support for revert.

According to the compatibility table, it was implemented in Safari first in 2016. It's indeed in that version's changelog. Maybe the rest of MSFNers will be delighted by such news. At least I find it refreshing. Mozilla only did it in 2019, followed by Google in 2020. Also means whooping 7 years passed since it was implemented in any browser first.

[UCyborg]

1 hour ago, AstroSkipper said:

FYI, the CSS userstyle https://userstyles.org/styles/180927/disable-css-animations-and-transitions-safely has unfortunately a negative impact in New Moon 28 on the download window which is not flexible anymore when this CSS style is activated.

Of course it does if you just copy pasted it without telling it to apply only to WWW, not browser chrome, like:

@-moz-document url-prefix("http://"), url-prefix("https://") {
  *, *:before, *:after {
    animation-delay: 0ms !important;
    animation-duration: 0.01ms !important;
    animation-iteration-count: 1 !important;
    transition-delay: 0ms !important;
    transition-duration: 0.01ms !important;
    /*transition: none !important;*/
  }
}

Edit: This should still be safer than the new prefs, eg. if you open the login form on MSFN, it doesn't hide when you click something else to hide it.

Edited September 16, 2023 by UCyborg

[AstroSkipper]

3 hours ago, UCyborg said:

Of course it does if you just copy pasted it without telling it to apply only to WWW, not browser chrome, like:

@-moz-document url-prefix("http://"), url-prefix("https://") {
  *, *:before, *:after {
    animation-delay: 0ms !important;
    animation-duration: 0.01ms !important;
    animation-iteration-count: 1 !important;
    transition-delay: 0ms !important;
    transition-duration: 0.01ms !important;
    /*transition: none !important;*/
  }
}

Edit: This should still be safer than the new prefs, eg. if you open the login form on MSFN, it doesn't hide when you click something else to hide it.

Thanks for the hint! That makes sense to me, of course. To be honest, I hadn't thought about it at all. Except for this user style, however, I have never used any before. I will modify my script immediately as suggested when I am back at my desktop computer. 

Edited September 16, 2023 by AstroSkipper
Update of content

[UCyborg]

On the other hand, if it can break the browser GUI, who can say there isn't a web site out there that doesn't break because of it? 

Edited September 16, 2023 by UCyborg

[AstroSkipper]

16 minutes ago, UCyborg said:

On the other hand, if it can break the browser GUI, who can say there isn't a web site out there that doesn't break because of it? 

If only a website breaks, then it doesn't matter for me and is much better than breaking the browser GUI. I definitely need an intact download window in New Moon 28. 

[PPeti66x]

@roytam1

Another problem found in NM27 (current version - 16.9.2023): wikipedia triggers an instant crash. Previous version (9.9.2023) is OK.

[UCyborg]

Anyone tried YouTube CPU Tamer by AnimationFrame? Supposed to reduce CPU usage on YouTube, though I observed the opposite. The newest Firefox I have ATM (110), there YouTube by itself also maxes out a single core when you scroll far enough into the comments section, but this doesn't seem to happen with Chromium based browser.

https://www.iloveimg.com/ also has issues, should be apparent after trying any tool offered there. So still a lot of problems remain even after recent boost in compatibility.

Edited September 16, 2023 by UCyborg

[AstroSkipper]

 

2 hours ago, AstroSkipper said:

3 hours ago, UCyborg said:

Of course it does if you just copy pasted it without telling it to apply only to WWW, not browser chrome, like:

@-moz-document url-prefix("http://"), url-prefix("https://") {
  *, *:before, *:after {
    animation-delay: 0ms !important;
    animation-duration: 0.01ms !important;
    animation-iteration-count: 1 !important;
    transition-delay: 0ms !important;
    transition-duration: 0.01ms !important;
    /*transition: none !important;*/
  }
}

Edit: This should still be safer than the new prefs, eg. if you open the login form on MSFN, it doesn't hide when you click something else to hide it.

Thanks for the hint! That makes sense to me, of course. To be honest, I hadn't thought about it at all. Except for this user style, however, I have never used any before. I will modify my script immediately as suggested when I am back at my desktop computer. 

I tested your modified CSS style in New Moon 28, and now, it works as it should.

2 hours ago, AstroSkipper said:

2 hours ago, UCyborg said:

On the other hand, if it can break the browser GUI, who can say there isn't a web site out there that doesn't break because of it? 

If only a website breaks, then it doesn't matter for me and is much better than breaking the browser GUI. I definitely need an intact download window in New Moon 28. 

The download window in New Moon 28 is no longer negatively affected when the CSS style disable-css-animations-and-transitions-safely is activated. So, I'll leave it activated as I did before. Thanks again!

Edited September 16, 2023 by AstroSkipper
Update of content