Firefox 24 - 52 for ME and 98

[schwups]

Yes, it happened here, too, on Serpent 55, but not on NewMoon (latest) and Iceape (round about one year old). Others not testet on the fast. Exception 406D1388 could still indicate the same problem. I'll check the address.

Requirements: The KexVista module isn't required as I expected. RoyTam already has removed APISet Stub dll entries in dependentlibs.list. Otherwise, the same requirements as with FF52 should apply.

.

[schwups]

It is RaiseException in Xul. Affected Serpent 52/55. I tried to replace it, but Serpent still crashes on this page.

Edited September 1, 2023 by schwups

[schwups]

Workaround: Set WebRTC for those pages to disabled or permanently switched off (media.peerconnection.enabled -> false). The extension QuickJava 2.1.2 is a nice helper here. You can disable WebRTC with one click only. NewMoon has no WebRTC implementation. The crash also doesn't occur with Javascript disabled.

[schwups]

Does the crash also occur on pages other than browser checkers? My impression is that only a few pages are affected. WebRTC has always been inconspicuous with Firefox until now and and has not led to such crashes, but of course there is no replaced RaiseException.

The replace in xul also works with NewMoon 27, but not before version 27.10.0.win32-git-20230513.  Now NewMoon 27 works, too, but not before version 27.10.0.win32-git-20230513. It runs with AzureContentBackend Cairo only, so a lot of icons are "very abstract" or unrecognizable and buttons have randomly coloured borders (compilation switch - skia disabled). There are PM Themes, like PMOpera 2.1.0 or Aeromoon 2.2.0, which at least improve the appearance. The palemoon process can continue to run in the background for a while after closing. And note to be expected issues like input in address bar doesn't work on second browser start anymore / Search Engine List is empty or browser window is not completely displayed (Remedy: Hardware acceleration "layers.acceleration.force-enabled -> true"). Disable OMTC (offmainthreadcomposition) without Hardware acceleration. If necessary disable peerconnection (WebRTC), too.

 

Edited October 17, 2023 by schwups

[woi]

Yes, sometime it crashed on MSFN when media.peerconnection.enabled (WebRTC) is set to true (but it's rare and it doesn't crash most of the time). But since I disabled it, most of the websites that crashes now works fine (including MSFN).

EDIT: Forgot to say this but "Tiktok.com" doesn't work too (with WebRTC enabled).

Edited September 4, 2023 by woi

[schwups]

A crash doesn't occur on MSFN, , whether with or without WebRTC here. Other than most, which ones don't work for you?

[jumper]

• RaiseException=v4
• RaiseException=>GetFileTime

This is a debugging api so there might be others--use mode Windows XP Debug.

I've added two RaiseException wrappers to Kexbases.26. The debug version currently displays the parameter values before passing on the call. The default version will ignore the call if it is "continuable", or call the debug version if not.

 

[schwups]

14 hours ago, jumper said:

• RaiseException=v4
• RaiseException=>GetFileTime

This is a debugging api so there might be others--use mode Windows XP Debug.

I've added two RaiseException wrappers to Kexbases.26. The debug version currently displays the parameter values before passing on the call. The default version will ignore the call if it is "continuable", or call the debug version if not.

 

Jumper, thanks for taking on the problem.

I also can start Serpent with following Kstub entries without replacing RaiseException in xul or mozjs:

[Kernel32.dll]

RaiseException=v4

or

RaiseException=>Kernel32.dll:GetLastError

However, I haven't tested this method in depth to see, if it causes any problems.

GetFileTime doesn't work for Serpent. What makes you think, that it's a good choice in this case of RaiseException?

Supplement: RaiseException=>api-ms-win-core-errorhandling-l1-1-0.dll:GetLastError (Wine 5 APISetDLL) works, too, but you usually don't have wine 5 APISetDlls on your machine. The ReactOS file realisation also may work, but I haven't a good experience with ROS APISetDlls, because some have produced errors. API SetLastError produces an Error on Serpent close.

Edited September 5, 2023 by schwups

[schwups]

v4 seems to be the best  for now. It even solves the crashes on browser checker sites, if WebRTC is enabled.

[schwups]

On 9/3/2023 at 5:00 AM, woi said:

EDIT: Forgot to say this but "Tiktok.com" doesn't work too (with WebRTC enabled).

v4 works here, too. Note, that the Kstub.ini with RaiseException in core.ini must placed on the right to Kexbasen,Kexbases. It doesn't work, if it is on the left side.

[woi]

Now since I added RaiseException=v4 on my Kstub824.ini, tiktok.com and whatismybrowser.com now works fine with WebRTC enabled. Also I replaced Serpent 52 modified mozjs.dll and xul.dll with the one that is unmodified.

 

Edited September 6, 2023 by woi

[woi]

54 minutes ago, schwups said:

Note, that the Kstub.ini with RaiseException in core.ini must placed on the right to Kexbasen,Kexbases. It doesn't work, if it is on the left side.

For me it looks like this:

contents=std,kexbasen,kexbases.kstub824,kexvista

Not sure if I'm on the right track or not.

Edited September 6, 2023 by woi

[schwups]

5 hours ago, woi said:

contents=std,kexbasen,kexbases.kstub824,kexvista

IMO the entry RaiseException=v4 is OK for testing purposes. For testing only, because such entry may cause errors elsewhere, e.g.  the first I found with v4 is MRU-Blaster hangs on close. As soon as we have the Kex update, RaiseException should then be removed from Kstub.ini at the latest.

Edited September 6, 2023 by schwups

[woi]

Also, there is some glitches at the title bar when using the Basilisk Photonic theme https://addons.basilisk-browser.org/addon/photonic/ (SP52). But it can be fixed by going to "Customize -> Title Bar to enable it.

It also happens on Windows 2000 (with Extended Kernel) as well, but it's not as bad as Windows 9x (no fading effect).

Edited September 16, 2023 by woi

[Kurt_Aust]

I've got a weird effect in that when I use the portable version of Firefox 38 (from the 1st post) it always crashes on exit but otherwise works fine.

However if I use the non-portable version that it calls and is packaged with, there is no crash on exit, is this expected behaviour or is there something else I have to do?