My Browser Builds (Part 4)

[Ugh]

Does anybody have both Firefox ESR 52.9.0esr and the likes of XP-compatible Pale Moon 26.5.0 recompiled without `-msse2` option, that is, it shall be non-SSE2/noSSE2 but only single SSE-compatible? The reason for this is two-fold: 1) Windows XP and 2) non-SSE2 CPU. Thanks in advance for sifting through your archives!

[roytam1]

2 hours ago, UCyborg said:

Downloading goes really slow, about 25 KB/s.

someone is trying to "mirror" my binaries.

[roytam1]

44 minutes ago, Ugh said:

Does anybody have both Firefox ESR 52.9.0esr and the likes of XP-compatible Pale Moon 26.5.0 recompiled without `-msse2` option, that is, it shall be non-SSE2/noSSE2 but only single SSE-compatible? The reason for this is two-fold: 1) Windows XP and 2) non-SSE2 CPU. Thanks in advance for sifting through your archives!

my PM26.5-win2000 builds are already noSSE.

and also my -ia32 builds are also noSSE (still needs CMOV instruction set anyway)

[Ugh]

5 hours ago, roytam1 said:

my PM26.5-win2000 builds are already noSSE.

and also my -ia32 builds are also noSSE (still needs CMOV instruction set anyway)

That was fast! I want to update a user's machine with the latest non-SSE2 (recompiled without `-msse2` option if it's feasible, but couldn't find the links) Firefox ESR 52.9.0esr if that would be possible, to avoid this old lady's needs to change her browsing habits. Any source?

Edited April 8, 2023 by Ugh

[Ugh]

Also it seems Pale Moon - Release Notes for Archived Versions has removed Windows XP support as of 27.0.0 (2016-11-22) version. Does it still hold the ground that for WinXP there is no newer PM-based build in the wild other than PM26.5 ? Thank you!

[RamonUn]

3 hours ago, Ugh said:

there is no newer PM-based build in the wild other than PM26.5 ? Thank you!

There are indeed no more official pale-moon builds for XP newer than 26.5 but the whole point of this thread is to talk about all the unofficial builds that are based on latest Pale Moon, and also PaleMoon27 for XP, And some of those builds are compatible with SSE1 cpu as well.

[Ugh]

12 hours ago, RamonUn said:

some of those builds are compatible with SSE1 cpu as well

Cool! That, while my ESR 52.9.0esr question stills here, was (is) the second part I had asked about. Can anybody link those in? Conditions are simple: XP & noSSE2.

Edited April 9, 2023 by Ugh

[VistaLover]

@Ugh: What does your "old lady" use currently, BTW?

Official FirefoxESR 52.9.x was never meant to be run on a non-SSE2 CPU; AFAIAA, no effort has been made public here to recompile it to run on a SSE-only machine; compiler optimisation is one thing, but the source-code itself is another ; if the latter invokes the SSE2+ instructions set, then the SSE-only optimised binary will crash when run on a SSE-only CPU...

The last (official) Firefox that would run on a SSE-only CPU is FxESR 45.9.x; Roytam1 does maintain a fork of that, for the sake of very old H/W, but it's being rarely updated and offered (as ALL of his browsers) in English, only:

http://o.rthost.win/gpc/files1.rt/firefox-45.9.34-20221210-b637891ad-win32-sse.7z

Discussion of Firefox 45/52 tends to be academic nowadays, because Google, with their dominance on everything web-related , have restructured the Internet of 2020+ to be palatable mostly to the latest Chrom-e/-ium version (-20 versions, if lucky ...) and likewise with their stepchild, "Firefox Browser"...

The "-sse" & "-ia32" variants of New Moon (27/)28 target nonSSE2 CPUs:

New Moon 27.10 was originally forked from Pale Moon 27 (Vista+); its engine is inadequate for today's web, but it has recent TLS (HTTPS) implementation...

https://o.rthost.win/palemoon/palemoon-27.10.0.win32-git-20230408-f4385096ea-xpmod-sse.7z

https://o.rthost.win/palemoon/palemoon-27.10.0.win32-git-20230408-f4385096ea-xpmod-ia32.7z

New Moon 28.10.6a1 tries to catch up/be better than latest Pale Moon 32.1.0 (Win7+) ; it's based on the UXP platform and has, together with Serpent 52, the best web compatibility of the offered XP-compatible browsers (but still inferior to mainline Chrome/Firefox): 

https://o.rthost.win/palemoon/palemoon-28.10.6a1.win32-git-20230408-d849524bd-uxp-b5e969b8a-xpmod-sse.7z

https://o.rthost.win/palemoon/palemoon-28.10.6a1.win32-git-20230408-d849524bd-uxp-b5e969b8a-xpmod-ia32.7z

Originally forked from FxESR 52, the "-ia32" variant of Serpent 52.9.0 will be the closest thing (to Fx52) that will run under XP+nonSSE2:

https://o.rthost.win/basilisk/basilisk52-g4.8.win32-git-20230408-3219d2d-uxp-b5e969b8a-xpmod-ia32.7z

 

ALL above binaries will NOT autoupdate, thus you need to manually update/overwrite older builds with newer ones (they're released in weekly cycles, but, unless you have a special need that calls for it, you can update more infrequently); and, as stated, are only offered in English (so, here's hoping your client has at least some rudimentary command of English ) ...

For updated builds, you may want to bookmark:

https://rtfreesoft.blogspot.com/search/label/browser

Best regards

Edited April 9, 2023 by VistaLover

[Mathwiz]

On 4/3/2023 at 5:17 PM, VistaLover said:

After a proper browser restart, multiprocess was ON, with a maximum of 3 "basilisk.exe" processes inside Task Manager ; I used it that way for more than an hour, no dramatic changes in my configuration here (3GB of total RAM, 2 core CPU, 2007-era...), the browser was rock-steady overall ; but then I stumbled on a show stopper ; once inside an online post editor, like the one here on MSFN (I'm typing this on) or GitHub's issue/comment editor, the DEL and BACKSPACE keyboard buttons no longer work as expected there

I must admit, I never tried GitHub's editor with multiprocess (m10s?) on; unlike you I rarely post anything there; besides, until recently, accessing GitHub required palefill, which is incompatible (i10e?) with m10s mode anyhow, so I had always opened GitHub links in a single-process profile instead.

The issue you describe, though, sounded just like a bug that plagued several recent versions of Serpent. So I wondered if the same workaround would work? The workaround for the recent bug was setting the pref dom.keyboardevent.keypress.dispatch_non_printable_in_content (d58t?) to true.

I had never reset that pref after the very recent fix, so I never realized there was still a problem while in m10s mode, but I can now confirm that resetting that pref to false (the default) does disable the normal functioning of BACKSPACE, and setting it to true makes BACKSPACE function normally again!

Edited April 9, 2023 by Mathwiz

[roytam1]

On 4/4/2023 at 12:15 AM, Saphir said:

Hmm.... Can we finally hope for a chance now that ECMAScript related Pale Moon "villains" are able to be implemented a bit more easier in the future?

https://repo.palemoon.org/MoonchildProductions/UXP/issues/2173 - JS: Parser frontend overhaul
https://repo.palemoon.org/MoonchildProductions/UXP/pulls/2189

Seems also someone wants to give a second try towards this "villain" here:
https://repo.palemoon.org/MoonchildProductions/UXP/issues/1691 - Dynamic Module impport

and also this: https://repo.palemoon.org/MoonchildProductions/UXP/issues/2158

this: https://repo.palemoon.org/MoonchildProductions/UXP/issues/2142

and this: https://repo.palemoon.org/MoonchildProductions/UXP/issues/2097

Edited April 9, 2023 by roytam1

[VistaLover]

13 hours ago, roytam1 said:

and this: https://repo.palemoon.org/MoonchildProductions/UXP/issues/2097

... Which I've already drawn attention to in my post of Apr 4th :

On 4/4/2023 at 12:11 AM, VistaLover said:

in my casual browsing I now find the most "mean" UXP villain to be nullish coalescing assignment ("??=") operator , for which UXP issue #2097 has been opened  ...

BTW, could you possibly offer some help towards:

On 4/4/2023 at 12:11 AM, VistaLover said:

I have a vested interest in opening

https://venomissimo.notion.site/venomissimo/FFmpeg-86-3b484982448b485eaed6b687b2f67047

https://venomissimo.notion.site/4473a6dcc8494218be42fc504f67d5e0?v=7acc381ce8ef4a16bb95f14db710588e

https://www.notion.so/34dc4ddf501a4b98b46ea9fb4f3470af?v=878345c5d88f4d21a6520db752b5c29f

in St52, but all these URLs end up in blank tabs currently, because "*.notion.(site|so)" use the despicable "??=" operator since mid-January ... Can a kind heart offer some solution/filter for "Modify HTTP Response", please?

Thanks in advance ...

OT: Happy Easter Sunday to all those that observe it today (it's on Apr 16th in my parts ...).

Edited April 9, 2023 by VistaLover

[Saphir]

14 hours ago, roytam1 said:

and also this: https://repo.palemoon.org/MoonchildProductions/UXP/issues/2158

this: https://repo.palemoon.org/MoonchildProductions/UXP/issues/2142

and this: https://repo.palemoon.org/MoonchildProductions/UXP/issues/2097

Happy Easter everyone!

One villain was finally fixed and resolution added also today and one other addition:

Add window.event #595=
https://repo.palemoon.org/MoonchildProductions/UXP/issues/595

Implement self.structuredClone() #2197
https://repo.palemoon.org/MoonchildProductions/UXP/issues/2197

-------------------------------------------------------------------

And for everyone who wants to follow the progress of:

Class Fields and Initializers #8
https://repo.palemoon.org/martok/UXP-contrib/issues/8

and

Implement dynamic module import #1691
https://repo.palemoon.org/dbsoft/UXP/src/branch/dynamic-module-import

Edited April 9, 2023 by Saphir

[UCyborg]

Memory management topic popped up on Pale Moon forums recently. I've always assumed the browser's codebase has memory leaks, but they say it's caching and by design. Strange design... And that thing about Gmail (or any site leaking memory) doesn't compute in my head. Last time I checked, you don't do mallocs and frees in JavaScript, so how can you "leak" memory as a website? Isn't that what garbage collector is for?

It seems this design can lead to browser eventually hanging, at least or especially 32-bit versions, presumably due to all gunk hanging around, plus extensions also pollute process' address space, can't tell whether the impact is significant. At least those that aren't unpacked, its XPI file is mapped in the process' address space.

So more RAM and running 64-bit version may prevent the gunk from killing the browser, but it won't prevent sluggishness that occurs over time, for which the only cure is restart of the browser. From my experience, I wouldn't say heavy "Googlized" sites alone are the factor.

[RamonUn]

I do have to restart UXP based browsers more and more often, at some point the cache has to be dropped otherwise it is the same thing than a memory leak.

It is not trivial for a programmer though to decide what is best to cache and for how long. I do remember an old Article from Raymond Chen taking about this and hw Acess was able to improve its performance by not caching some results that were easy to recalculate. For sure UXP has a problem and there is a lot of cache they should drop and that never gets dropped, otherwise it would not seems normal to everyone that a session is supposed to get slower and slower. and that you hould just upgrad your RAM.

https://devblogs.microsoft.com/oldnewthing/20041220-00/?p=36943

[Saphir]

46 minutes ago, UCyborg said:

Memory management topic popped up on Pale Moon forums recently. I've always assumed the browser's codebase has memory leaks, but they say it's caching and by design.

Well, one site which instantly slows down when you had it open is Outlook - Do that once and the browser goes down instantly, slowdowns appearing.

So yeah, it has some serious memory issues, the question is how to solve them.