My Browser Builds (Part 4)

[Ben Markson]

7 hours ago, VistaLover said:

If you want it to always be there, no matter how narrow an unpinned tab is, then this applies

Setting browser.tabs.tabClipWidth to zero is the answer. I don't normally open so many tabs that it makes a difference and only noticed because I was experimenting with the recent tab changes. Anyway, mystery solved. Thank you.

Ben.

Edited March 12, 2023 by Ben Markson
typo

[modnar]

A rather funny Serpent 52.9 crash just happened - I was watching youtube for some time and then HWunBoxed video RTX4070Ti against RX7900XT and the message about basilisk encountering error popped up.

The video was still going for a while and then stopped and if I dragged the error window all over the screen (which I did) resulted it this lovely picture (attached).

When I "click(ed) here" it said it has to do with mozglue.dll or something similar.

BTW thank you for the new release, roytam1, starting from last week's each now is much more up to date and faster working than ever before.

After a Serpent restart it works normally again.

Edited March 12, 2023 by modnar

[AstroSkipper]

3 hours ago, modnar said:

A rather funny Serpent 52.9 crash just happened - I was watching youtube for some time and then HWunBoxed video RTX4070Ti against RX7900XT and the message about basilisk encountering error popped up.

The video was still going for a while and then stopped and if I dragged the error window all over the screen (which I did) resulted it this lovely picture (attached).

When I "click(ed) here" it said it has to do with mozglue.dll or something similar.

BTW thank you for the new release, roytam1, starting from last week's each now is much more up to date and faster working than ever before.

After a Serpent restart it works normally again.

What a coincidence! I just had a browser crash regarding mozglue.dll in the very latest release of New Moon 28 (28.10.6a1 (32-bit) (2023-03-10)). Here is the entry from the event log:

Ereignistyp:	Fehler
Ereignisquelle:	Application Error
Ereigniskategorie:	Keine
Ereigniskennung:	1000
Datum:		12.03.2023
Zeit:		16:15:32
Benutzer:		Nicht zutreffend
Computer:	
Beschreibung:
Fehlgeschlagene Anwendung palemoon.exe, Version 4.8.6.8297, fehlgeschlagenes Modul mozglue.dll, Version 4.8.6.8297, Fehleradresse 0x00004cd1.

Weitere Informationen über die Hilfe- und Supportdienste erhalten Sie unter http://go.microsoft.com/fwlink/events.asp.
Daten:
0000: 41 70 70 6c 69 63 61 74   Applicat
0008: 69 6f 6e 20 46 61 69 6c   ion Fail
0010: 75 72 65 20 20 70 61 6c   ure  pal
0018: 65 6d 6f 6f 6e 2e 65 78   emoon.ex
0020: 65 20 34 2e 38 2e 36 2e   e 4.8.6.
0028: 38 32 39 37 20 69 6e 20   8297 in 
0030: 6d 6f 7a 67 6c 75 65 2e   mozglue.
0038: 64 6c 6c 20 34 2e 38 2e   dll 4.8.
0040: 36 2e 38 32 39 37 20 61   6.8297 a
0048: 74 20 6f 66 66 73 65 74   t offset
0050: 20 30 30 30 30 34 63 64    00004cd
0058: 31 0d 0a                  1..     

No idea what caused it! 

Edited March 12, 2023 by AstroSkipper

[VistaLover]

... Reading a certain MSFN post , it contained a link to below MS support article:

https://support.microsoft.com/en-us/topic/microsoft-net-framework-repair-tool-is-available-942a01e3-5b8b-7abb-c166-c34a2f4b612a

My (main) browser is (latest) St52; the page requested did not render properly :

I toyed a bit with disabling recently implemented WebComponents support and restarting the browser, same thing ... Tried Safe Mode, a fresh St52 profile, both to no avail ...

Upon further troubleshooting, it emerged ALL UXP-based browsers (+St55) suffer from this "ailment" ...
Below is latest NM28 in an almost fresh profile, when displaying:

https://support.microsoft.com/en-us/topic/update-to-enable-tls-1-1-and-tls-1-2-as-default-secure-protocols-in-winhttp-in-windows-c4bd73d2-31d7-761e-0178-11268bb10392

OTOH, the very OLD 360EEv11 (Ch69-based) has no issue displaying that URL properly:

My troubleshooting skills are failing me today , I couldn't get anything meaningful from St52's WebConsole...
@roytam1: What is exactly the problem here? Since Ch69 renders "support.microsoft.com" properly, surely it can't be something new and exotic devised by Google that causes the breakage in UXP ...

Workaround: While dealing with this, it emerged that my St52 copy can, indeed, render that site OK, provided I disable the JS served; I just used uBlock Origin for convenience, disabling "scripting" selectively on that site alone:

Plea for experts' help: I'm not that proficient in uBO's advanced features, can someone among you pinpoint the exact script responsible for the breakage? I spent ca. 30min with uBO's logger trying to identify it, but was not successful (not myself today ...) ; the logic here being to selectively block that script with a custom filter, instead of a blanket "no-scripting" rule on "support.microsoft.com"...

Thanks in advance ...

[UCyborg]

1 hour ago, VistaLover said:

@roytam1: What is exactly the problem here? Since Ch69 renders "support.microsoft.com" properly, surely it can't be something new and exotic devised by Google that causes the breakage in UXP ...

The (partial?) answer is on Pale Moon forum - UXP doesn't support <link rel="preload" ...>.

5 hours ago, AstroSkipper said:

What a coincidence! I just had a browser crash regarding mozglue.dll in the very latest release of New Moon 28 (28.10.6a1 (32-bit) (2023-03-10)).

Unsure if above is related to current work-in-progress on Web Components, but here's a good URL to crash right away in xul.dll: https://www.cbsnews.com/news/asteroid-2023dw-could-hit-earth-valentines-day-2046-nasa/?ftag=CNM-00-10aab7e&linkId=204708560

Should be fixed already by upstream.

Edited March 12, 2023 by UCyborg

[Mathwiz]

54 minutes ago, UCyborg said:

The (partial?) answer is on Pale Moon forum - UXP doesn't support <link rel="preload" ...>.

Pull-quote from another thread about a different site with this problem (https://forum.palemoon.org/viewtopic.php?f=70&t=26061&p=207349#p207323):

Quote

Looks like css is simply missing because they are using "link rel=preload" which is an experimental feature that isn't supported by us (even Firefox only JUST has it since 85, according to MDN, but of course it's a long-since implemented Chrome feature where this originated)

Emphasis added. And further down the page:

Quote

The fix is simple though, just use rel="stylesheet" instead of "rel="preload", remove the onload handler from the <link> elements (which requires JS for it to work, as well, by the way, which isn't good or necessary to just load stylesheets!), and done.

So it looks like Google gets the blame after all. They gave sites a needlessly complicated way to do something simple for no benefit (unless they consider breaking older Firefox-based browsers a "benefit")

[roytam1]

6 hours ago, AstroSkipper said:

I just had a browser crash regarding mozglue.dll in the very latest release of New Moon 28 (28.10.6a1 (32-bit) (2023-03-10)). Here is the entry from the event log:

this is an old problem since feb uncovered by https://github.com/roytam1/UXP/commit/9824659d3c6b1c4fdc2616f789f0696d1cbe2ef8 and https://github.com/roytam1/UXP/commit/b7e4530861fcc971aee9d867b3b376e31263486c and they said it is fixed by https://github.com/roytam1/UXP/commit/1f0df8f421626bcbc397512773f3043d72d96a4f but I still need to revert first 2 mentioned commits in order to make it not crashing.

[VistaLover]

2 hours ago, UCyborg said:

UXP doesn't support <link rel="preload" ...>

3 hours ago, VistaLover said:

Since Ch69 renders "support.microsoft.com" properly, surely it can't be something new and exotic devised by Google that causes the breakage in UXP ...

Many thanks @UCyborg , much appreciated! So, it wasn't actually "new and exotic", just "old and exotic (Googlism)" ...

1 hour ago, Mathwiz said:

(even Firefox only JUST has it since 85, according to MDN

Interestingly/funnily enough, Mozilla were quick to concoct their own implementation in Fx 56.0, which turned out to be problematic and had to be retracted in v57.0: 

1 hour ago, Mathwiz said:

but of course it's a long-since implemented Chrome feature where this originated

... Yes, as early as version 50 (!) of Chromium  ...

1 hour ago, Mathwiz said:

So it looks like Google gets the blame after all. They gave sites a needlessly complicated way to do something simple for no benefit (unless they consider breaking older Firefox-based browsers a "benefit")

... What I still don't get (being a non-coder) is why disabling JS works around it - and any answer to my uBO-related "plea-for-help", please  ?

[NotHereToPlayGames]

12 minutes ago, VistaLover said:

Yes, as early as version 50 (!) of Chromium  ...

But what I still cannot grasp, as far as timelines go, did Google "create" this Googlism, or were they just the first to be able to "render" what a THIRD-PARTY "CONSORTIUM" created?

 

edit - my frame of reference is the automotive industry.  here in the US, it is often Federal Mandates that come FIRST, then YEARS later is the "widget" created to meet that Mandate.

Edited March 13, 2023 by NotHereToPlayGames

[nicolaasjan]

8 hours ago, VistaLover said:

and any answer to my uBO-related "plea-for-help", please  ?

Well, I troubleshooted it with the help of eMatrix.

It appears there are 2 scripts from 'support.microsoft.com' that can be blocked (no idea what impact that has on functionality).

[Edit]

Checked a while later and there was only 1 script (according to eMatrix)...

But checking uBOs logger, I see a lot of scripts from support.microsoft.com/js/*.

Blocking all of these in uBO ---> page not functioning.

Conclusion: I'm no good at troubleshooting either.

Edited March 13, 2023 by nicolaasjan

[AstroSkipper]

17 hours ago, VistaLover said:

... Reading a certain MSFN post , it contained a link to below MS support article:

https://support.microsoft.com/en-us/topic/microsoft-net-framework-repair-tool-is-available-942a01e3-5b8b-7abb-c166-c34a2f4b612a

My (main) browser is (latest) St52; the page requested did not render properly :

For me, as a simple, temporary workaround, eMatix is helpful. In eMatrix, I blocked 17 scripts, and the website "Microsoft .NET Framework Repair Tool is available" becomes readable again in New Moon 28. Additionally, uBlock Origin (used in normal mode) is also enabled on this site. Here is a screenshot taken from my New Moon:



All is completely readable again.  I don't really care which functionality is broken by blocking these scripts. The main thing is that you can read the Microsoft articles again. I reloaded this website a few times and restarted my browser. The website is rendering without any problems and looks like always (when everything was fine). The download of the Microsoft .NET Framework Repair Tool works, too. A positive side-effect is this website is now easy to load and very responsive if blocking these scripts. No negative CPU impact anymore. Hope it stays that way for a while now.

Kind regards, AstroSkipper

Edited March 13, 2023 by AstroSkipper
Update of content

[legacyfan]

On 3/10/2023 at 4:36 PM, roytam1 said:

New regular/weekly KM-Goanna release:
https://o.rthost.win/kmeleon/KM76.4.7-Goanna-20230311.7z

Changelog:

Out-of-tree changes:
* update Goanna3 to git 35448d94ad...33f0551ea5:
- lineContainer can be NULL in nsIFrame::InlinePrefISizeData::ForceBreak(), workaround it to fix crash when loading https://football.kulichki.net/england/ (40afff632d)
- import changes from `dev' branch of rmottola/Arctic-Fox:
 - Bug 599328 - Clicking a label shouldn't draw a focus ring if clicking the control wouldn't have. r=enndeakin (fe391aecaa)
 - Bug 1221947 - part 1: add BYTOUCH flag to nsIFocusManager, r=jaws,smaug IGNORE IDL (41b191e6f1)
 - Bug 1192573 - Require tablet-mode (Win10+) to show the on-screen keyboard. r=masayuki (9244e1d044)
 - Bug 1197722 - Respect Win10 pref about bringing up the touch keyboard outside of tablet mode. r=masayuki (00e20db670)
 - Bug 1210033 - Add on-screen keyboard diagnostic information to about:support. r=masayuki r=felipe (0101168557)
 - Bug 1211805 - rework keyboard detection for on-screen keyboard, r=jaws (d6eee8e170)
 - Bug 1221947 - part 2: add TOUCH cause to IME Handling and use it to avoid checking for a keyboard, r=masayuki (b39ad48728)
 - Bug 1221947 - part 3: pass touch information to focus manager in RenderFrameParent, r=kats (84bbf77326)
 - Bug 1229258: Implement dummy asmMergeWith/retargetWithOffset functions in none backend; r=h4writer (cb7c19dbfa)
 - Bug 1210577. Remove useless virtual keyword from ScrollFrameHelper::ComputeScrollClip. r=mstange (5c40b8bd98)
 - Bug 1177505 - Make all continuations fluid within non-bidi paragraphs. r=smontagu (6d7956c20f)
 - Bug 1216096: restore previous RTL caret behaviour by backout of bug 1164963, bug 1177505, and bug 1180417. r=jfkthame (11d974c609)
 - Bug 1157752 - In vertical writing modes, text-orientation:upright should force LTR directionality. r=smontagu (b8843ce65e)
 - Bug 1002466 - Use SK_ScalarNearlyZero instead of 0.0f to render dots r=mattwoodrow (79406a8ce9)
 - Bug 1214309 - nudge zero dash intervals to be non-zero under Skia. r=jmuizelaar (b781a0c63a)
 - Bug 1208234: Stop the null pointer crash. r=bas (e7fb7f56ad)
 - Bug 1202317. Bail out early if RequestRect is going to do redundant work. r=mstange (cc85741516)
 - Bug 1167026 - Avoid division by zero when flattening a bezier curve segment with equal control points. r=bas (11e036081a)
 - Bug 1166879. Avoid uninitialized read in FlattenBezier. r=mstange (893177de38) (f3497a0291)
- import changes from `dev' branch of rmottola/Arctic-Fox:
 - Bug 1193786 - crash in mozilla::dom::Element::FindAttrValueIn, r=marcoz (1f03b9594f)
 - bug 1172516 - fix firing caret move events for proxied accessibles r=lsocks (6a62915a27)
 - Bug 582024 - ARIA active-descendant should work for ARIA owned elements, r=yzen (4fc7049fe1)
 - Bug 1212457 - crash at ARIARowAccessible::GroupPosition(), r=marcoz (dd058e3dcd)
 - bit of 1139049 (ae765adc6a)
 - Bug 1221542, bug 1221543 - crash in mozilla::a11y::DocAccessible::SeizeChild/PutChildrenBack, r=davidb (8a35b14523)
 - Bug 1105611 - HyperTextAccessible should set DOM range outside of pseudo elements r=surkov (19e2cf65cd)
 - Bug 1177765 - Add xmlroles for MathML. r=surkov (381201ae27)
 - Bug 1176683 - crash in mozilla::a11y::HyperTextAccessible::LandmarkRole() const, r=marcoz (f3fd977d35)
 - Bug 1176123 - Add NODE_CHILD_OF/NODE_PARENT_OF relations to mroot. r=surkov (36f08faa15)
 - Bug 1179483 - Fix crash in mozilla::a11y::HyperTextAccessible::RelationByType. r=MarcoZ (fec7fde5b3)
 - Bug 1177765 - Make nsIMathMLFrame expose the fence and separator properties of operators. r=karlt (91e45d9980)
 - Bug 1139709. Cache nsMathMLContainerFrame's intrinsic width. r=mats (ac40d07d89)
 - code style (c5f50fc0c3)
 - Bug 1226875 - Remove nsIFrame::GetLastChild(). r=mats (9b88566b77)
 - pointer style (6a5b9599f4)
 - Bug 1177093 - mathfont.properties does not need to be preprocessed after Bug 1000745 r=fred.wang (d114e285a1)
 - Bug 1224951 - Part 2: Fix -Wunreachable-code warnings in layout. r=dholbert (5854ee5e3e)
 - Bug 1141443 - Remove unused rpcns4 from OS_LIBS. r=mshal (1a5ccc985b)
 - bug 1218762 - proxy ia2Accessible::scrollTo{,Point}() r=davidb (db1db8e703)
 - Bug 1182208 - Add support for android scrolling and range accessibility actions. r=mfinkle r=yzen (29df38ee29)
 - Bug 1144516 - Remove offset of mozbrowser iframe. r=yzen (c7473a8feb)
 - Bug 1176292 - Send 'toggle-shade' control event to content on 3 finger triple tap. r=yzen (536932619a)
 - Bug 1182222 - Make Layerview support accessibility HTML navigation. r=yzen r=mfinkle (2beb411989)
 - Bug 1203697 - Add braille navigation. r=yzen r=mfinkle (d5a513e79e)
 - Bug 1209054 - Make 2 finger tap toggle pause speech. r=yzen (28bc8c16a6)
 - Bug 1214398 - Add highlight box to doc body if it is available and set z-index. r=yzen (9cca205849)
 - Bug 1182214 - Update highlight rect as you scroll. r=yzen (6d2738366a)
 - Bug 1182214 - Follow-up to fix Presentation.displayedAccessibles getter. r=me CLOSED TREE (6dd8ec3652)
 - Bug 1212528 - ensuring first seen document in ancestry is announced first. r=marcoz (fb8e9bc406)
 - Bug 1217038 - Remove for-each and legacy array comprehension from accessible/. r=yzen (810590b5d3)
 - Bug 1220860 - fixing an error with getting documentElement on AccessFu stop. r=eeejay (4a35fabc3d)
 - Bug 1169019 - Removed screenreader announcement from gecko. r=yzen (2b1732cddc)
 - Bug 1211122 - ensuring that we check position against an actual doc/dialog on doc load event. r=marcoz (78b7cc8a77)
 - Bug 1019432 - [AccessFu] Only capture explore by touch events in Android. r=yzen (eee4b09c72)
 - Bug 1182311 - Make 3 finger triple tap more reliable. r=yzen (7cf926a839)
 - Bug 1201146 - Introduce "Section" traversal rule. r=yzen (253d53ec8d)
 - Bug 1203283 - Introduce Android "Control" traversal rule. r=yzen (4d14ffa973) (e0e84a7434)
- import changes from `dev' branch of rmottola/Arctic-Fox:
 - Bug 1216851 - Allow flattening of opacity when it contains text. r=roc (f62dead4b9)
 - Bug 1219833 - Respect composition underline color; r=masayuki nsTextFrame didn't use the composition underline color if the composition didn't have a foreground color defined. This patch makes it use the underline color if foreground color is not defined. (02bcedf5fb)
 - Bug 1211085 - Disable global reuse on b2gdroid r=khuey (a4bfc7c4e1)
 - Bug 1206817 - Take ownership of error reporting in the component loader. r=bz (d8b9b02dfb) (40ee015dc8)
- import changes from `dev' branch of rmottola/Arctic-Fox:
 - Bug 1172382 - Enable AccessibleCaret on B2G. r=roc (f5c58c2798)
 - remove duplicated (8823cc4a43)
 - Bug 1219310 - part 1 - ask the prefs file for its size directly; r=njn (93073cbc5e)
 - Bug 1219310 - part 2 - keep track of how much pref file we have read; r=njn (6a2a10a8b1)
 - Bug 1213123 - Make Preferences::SetString accept char16ptr_t instead of char16_t*. r=froydnj (a895a36861)
 - Bug 1216901 - Make the FasterMake backend reticulate splines when moz.build or jar.mn files are modified. r=mshal (ed4ec93b18)
 - Bug 1219122 - Move webapprt.ini definition to moz.build. - Add corresponding webpprt files. (da6bc91b5c)
 - Bug 1219126 - Move greprefs.js definition in moz.build. r=mshal (52f404c935)
 - code style (1b1e543834)
 - Bug 1162690 - Remove malformed uri warning in nsURLParser::ParseURL r=mcmanus (8534fcebb7)
 - Bug 1163028 - URL: stop escaping [ and ] in path r=mcmanus (f2f3deec40)
 - Bug 1163030 - URL: stop escaping ` in query r=mcmanus (17d6c07640)
 - Bug 1191423 - Disallow illegal characters in cookies set via HTTP. r=jduell (b1786d140f)
 - Bug 1210235 - Skip package verification if pref out or no signature. The package would be treated unsigned. r=valentin (63870dd7ef)
 - Bug 1216062 - Notify OnStartSignedPackagedRequest with package identifier. r=valentin. (81a14af3db)
 - Bug 1214079 - Doom the package cache if the signature verification failed. r=valentin (83824c2d5d)
 - Bug 1178448 - Use imported CA in developer mode. r=keeler,valentin (b9cf64b477)
 - Bug 1216469 - Bypass verification for signed packages from trust origins. r=valentin (a36d0a6d2f)
 - Bug 1218284 - Match signed packages' with trust origin without suffix. r=valentin (45529dc7df)
 - Bug 412457 - should unescape hostname first, then perform IDNA r=mcmanus (23ebe47574)
 - Bug 1217316 - Remove for-each from netwerk/. r=jduell (8d0ca69e9e)
 - Bug 1208847 - Add telemetry to measure how often secure cookies are set from non-secure origins r=mcmanus (57ecf3651d)
 - Bug 1165267 - Part 1: Replace appId and inBrowser by originAttributes v2. r=honzab (7710301407)
 - Bug 1165267 - Fix downgrading issue by restoring appId and inBrowserElement columns v3. r=honzab (3e8b8e4dfb)
 - Bug 1221049 - Use originAttributes from TabContext. r=kanru (5eaebe3b28)
 - Bug 1197944 - Change pref so that the http auth dialog is presented for sub resources as well. r=jduell (e3a7e2a1a7)
 - Bug 1202421 - Rename the network.auth.allow-subresource-auth pref. r=michal (87e29e1fdf)
 - Bug 1213577 - Use OriginAttributes in nsHttpAuthManager, r=mcmanus (33d0a25ac4)
 - Bug 961049 - Part 1: Remove unused members and methods; r=baku (0f3e6de06b) (33f0551ea5)

* 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

great job again on your excellent work your the best!

[nicolaasjan]

2 hours ago, AstroSkipper said:

For me, as a simple, temporary workaround, eMatix is helpful. In eMatrix, I blocked 17 scripts

Now I see why you have way more scripts from support.microsoft.com and other stuff at the top.

You don't have "Resolve CNAME records" checked in eMatrix.

Edited March 13, 2023 by nicolaasjan

[VistaLover]

6 hours ago, AstroSkipper said:

this website is now easy to load and very responsive if blocking these scripts.

... FWIW, with JS disabled, the top drop-down menus are "unresponsive" (and with it enabled, their content becomes just links) : 

[VistaLover]

23 hours ago, UCyborg said:

here's a good URL to crash right away in xul.dll:

https://www.cbsnews.com/news/asteroid-2023dw-could-hit-earth-valentines-day-2046-nasa/?ftag=CNM-00-10aab7e&linkId=204708560

... FWIW, it doesn't cause an instant crash if on previous Serpent v52.9.0 (2023-03-02) (32-bit); like my friend (I hope ) @NotHereToPlayGames wrote in the past: "Latest isn't necessarily greatest"; but that's to be expected in St52, on a de facto "unstable/weekly" distribution "channel" ...

23 hours ago, UCyborg said:

Should be fixed already by upstream

Actually, this issue was discussed in the official PM Forum; their fix did not make it into last weekend's UXP releases by a matter of just a few hours, but has since been ported to Roy's UXP tree ...

To add to that, I have been suffering myself from instant crashes in xul.dll (presumably attributed to the same root cause ) when the below combination is met:
a) a WE userscript manager is installed and enabled (in St52), such as Violentmonkey
b) the GitHub download zip userscript has been installed and enabled
c) when browsing GitHub in general, you (unexpectedly) arrive to a GH 404 page, e.g.

https://github.com/ytdl-org/youtube-dl/issues/31950

When on GH, I need that userscript enabled, so I've now gone back to previous St52 build (I don't use TMP), because I spend lots of time there and one never knows when a GH 404 page may come up on you ; hopefully, next weekend's build will relieve me from those crashes ...