Pale Moon x86-64 SSE2/AVX2

[Kris_88]

don't know what Kris's problem is but that person needs to think about eventually replacing old hardware & upgrading
...
anyone staying with such old hardware for more than 15 years is really stuck living in the past.

You are not right.

https://www.asrock.com/mb/Intel/J5040-I ... asp#Manual
https://www.intel.com/content/www/us/en ... tions.html

... and did it myself.
And 5 more that i don't really need.
All it costs me is odd 2 hours of machine time that doesn't even lock me out from using computer.

So it's really easy and you generate 6 versions in 2 hours.
Then it is even more unclear what prevents Moonchild from spending 20 minutes and making one version for compatibility. For the compatibility that he himself broke with his decision. And considering the fact that he has a ready-made environment for generation.

So again, what is preventing you from doing what they do?

The browser is used almost constantly and therefore high security requirements are imposed on it. As has already been discussed on the forum, there is no way to verify that the executable module is built from the sources that are presented in the repository, that malicious code or special vulnerabilities were not artificially embedded into these sources before building. Obviously, Moonchild does not check this. Thus, the problem comes down to trust in the person who performs the build. I have my own set of evaluation criteria for trust. One of them is the presence of an official certificate and signature of executable modules.

[UCyborg]

don't know what Kris's problem is but that person needs to think about eventually replacing old hardware & upgrading (heck, make & save money to get new computer; at least I have the money and the GUTS to do that). anyone staying with such old hardware for more than 15 years is really stuck living in the past.

Off-topic:
The past was nicer to live in. Firefox 4 runs circles around any modern browser, also there was no Cloudflare to worry about. And you didn't need a smartphone (or anything Android) to access online bank, which 99% of them require these days!

The stupid bank where I have the account dropped Android 7 in its app last December, so probably I won't even be able to login on the website at some point due to stupid 2FA authentication. They replaced the web browser certificate years ago, which worked perfectly, with a stupid retarded app.

I've been looking at new smartphones for two weeks now and I'm just lost and running in circles. Thin almost non-existent bezels with the stupid selfie camera drilled at the top of the screen (who drills camera literally IN THE BLOODY SCREEN!?), no notification LED and flickering OLED screens everywhere. The lower the brightness, the worse it is, and my eyes prefer lower brightness (no dark mode, though, please). Some models I've been looking at even have reports of static imagery burn-in. It's like we're back at the crappy CRTs, I hated those when I was a kid.

The 10 years old Sony Xperia E3 with 1 GB of RAM running LineageOS 14.1 still makes calls, sends SMS and can access shitty modern websites with Chromium 119 (wrapped in Via browser, which makes resource consumption acceptable), including those that want stupid WASM SIMD for some retarded reason, thankfully, the latter are not the norm yet. Some particularly bloated sites are problematic, but mostly, it's possible to get by.

Back to computers, I still can't find the good reason to replace my working computer equipment. The computer I'm writing this from will be 16 years old this April. The main LG screen will also be 16 sometime this year, the secondary Samsung will be 21. Maybe if I played modern demanding games or do anything else very demanding on the computer, I'd think otherwise.

Unlike computer at workplace (circa 2018, maybe slightly older), Windows Event Viewer doesn't contain an endless spam of errors regarding broken TPM module (no TPM in home computer). Also both Samsung screens at work, again much newer than ones at home, have some glitches, one flickers for a while when turned on, other shows one or two green vertical lines (which then disappear, but still). The screens at home have zero flickering, zero green lines and zero dead pixels, despite their age.

Back in 2023, I had to replace old LG smart TV at home, the TV broke because it fell to the floor in the middle of the night due to the stand giving up under TV's weight (seriously!). I was at the shop with my family, didn't consider researching beforehand, a TV's a TV, right? Wrong. I picked one of the Samsung models, it was the last one available as it was actually current previous year. Who would thought viewing angles would be so terrible that things go grey when looked at slightly from the side? And the ridiculously minimalist remote with hardly any buttons, despite there being plenty of space.

So me being discontent with modern tech is an understatement. It's not about the money either, I could buy few high-end computers if I wanted. But money doesn't buy you happiness and new computer won't enrich my life. And it seems pointless to buy one just for Pale Moon.

I don't have a problem with contributed 64-bit SSE2 build though and I could probably compile one myself in emergency. Even switching to 64-bit build was rather recent for me, the 32-bit (where official build still doesn't impose AVX) might still fare if you keep low amount of open tabs and avoid memory leak inducing sites (though easier said than done).

Kris_88 might have stricter requirements regarding certificates and what not. I could too, but it seems a bit hypocritical considering the risks I take in other areas of life. While it could happen that Nuck-TH turns evil and adds malware into executables or his environment being compromised and published executables somehow ending up infected, it seems about as likely as my car catching fire while on the road. Hey, it could, it has a tank full of flammable liquid in the back.

Off-topic:
In the end, it all comes down to majority of modern software being bloated, inefficient and imposing requirements for infinite amount of gigabytes and gigahertz for accomplishing the same things we've being doing a decade ago. I have the app of the local gas station on my phone. Paying for gas or a coffee on the coffee machine via app only requires processing small amount of data. So why does the app need over half minute to load? It sometimes fails to load and outputs an error and has to be re-launched. Why is quad-core from 2014 not enough?

A bit of offtopic, apropos 'one can do it'. I see in the Building instructions (Windows-based) the requirement for a Windows machine with Visual Studio, and a lot of GCC-style compiler options in the `.mozconfig`. Are those processed by (modern) Visual Studio? Me, I've been imagining the Windows builds are produced with MINGW.

AFAIK, non-Microsoft compilers weren't considered to be of sufficient quality for compiling the program of this caliber on Windows. Or something along those lines.

[Moonchild]

it is even more unclear what prevents Moonchild from spending 20 minutes and making one version for compatibility.

Adding a version doesn't stop with building it. It also involves release engineering, and not in the least the support needed for publishing multiple versions and the resulting ambiguity what our system requirements are.
Adding a version to my setup for you, would also mean I'd have to consider other non-standard environments, like Windows Vista, XP-with-kernel-hacks, Weird Atom cpus or SoCs, people in need of special versions for individual workflows, special versions to counter local surveillance, I could go on.

there is no way to verify that the executable module is built from the sources that are presented in the repository

And you don't have that guarantee from my built binaries, either. If that is a concern, you should build from source yourself (like is routinely done in high-sec environments).

On one hand you dismiss the effort involved in building and publishing a version that suits you. On the other hand you don't do so when it is asked you build yourself if you can't trust Nuck-TH but somehow can trust me (why?). Do you not see the dichotomy there?

AFAIK, non-Microsoft compilers weren't considered to be of sufficient quality for compiling the program of this caliber on Windows. Or something along those lines.

We are using MSVC for Windows builds. It most definitely has sufficient quality for this, even though we are often pushing the limits of any compiler in use. The resulting binaries are also more reliable and more performant than those generated by gcc-on-windows. While we use a unix-like shell environment to build, the actual compiler and linker are those of Visual Studio (i.e. MSVC).

[Kris_88]

Adding a version to my setup for you, would also mean I'd have to consider other non-standard environments, like Windows Vista, XP-with-kernel-hacks, Weird Atom cpus or SoCs, people in need of special versions for individual workflows, special versions to counter local surveillance, I could go on.

That is, you assume that you will be overwhelmed with demands to build a browser for all possible platforms ... However, there is an objection that this is not some other platform. This is the same Windows 8-11 on the same Intel processor.

On one hand you dismiss the effort involved in building and publishing a version that suits you. On the other hand you don't do so when it is asked you build yourself if you can't trust Nuck-TH but somehow can trust me (why?). Do you not see the dichotomy there?

Building a browser yourself is really not that hard, if you have a prepared environment. You have one, but many users do not (including me). And creating the necessary environment is much more difficult. I'm not even sure that VS2022 will work on Windows 8.1 (which I use). I probably need to install Windows 10 first, at least...

Regarding trust, I already wrote here:
viewtopic.php?f=5&t=31401&start=60#p254515

...
And we come to the conclusion that everything depends on trust in a specific person.
For example, quite a lot is known about Moonchild. He has been working on this project for a long time, he is the owner of the project, he is the owner of the trademark, he has a real (not self-signed) certificate, his first name, last name, place of residence, and approximately his age are known. And although each piece of information does not prove or guarantee anything, but in total, I personally have enough information to trust Moonchild.

As for the authors of third-party builds, I know much less about them.
And therefore, as I already said, there is a big difference between official builds and third-party builds.

[Moonchild]

Still missing my points. Never mind, I guess.

[back2themoon]

Off-topic:
This is getting worse than Cloudflare.

Kris_88, this topic is about Nuck-TH own builds which you clearly don't trust, as you've implied a million times already. So, why keep posting here?

This is the part where a normal person would suggest you to open another topic, and make your request there. Problem is, you have repeated the same request a gazillion times over and got the same answers all over again. When will it be enough? (when you get your candy I suppose).

Again (and again), please use your talents in a more fruitful way instead of endlessly repeating yourself.

[Kris_88]

Off-topic:

So, why keep posting here?

It's hard for me not to respond when someone addresses me directly. Besides, I don't think I went off-topic. Maybe just a little. But in general, everything I wrote here is somehow related to the build from this topic.

And you take on the role of moderator (even though you are not a moderator or the owner of the topic)
I think I've seen something like this before...

[Moonchild]

Off-topic:
If you want to further discuss this, Kris_88 and others, open a new topic in the browser development board.
That way, this topic can remain focused on updates and releases of the AVX2/SSE2 builds of the browser with minimal distractions.

[back2themoon]

Off-topic:

It's hard for me not to respond when someone addresses me directly... ... And you take on the role of moderator (even though you are not a moderator or the owner of the topic)
I think I've seen something like this before...

You didn't just respond, you repeated almost everything. Again.

As for the moderator thing (a trendy accusation, it seems): I am only reminding you, as a member and subscriber of this topic that you keep hijacking it with the exact same request. Did it ever occur to you that subscribers of this topic might not be interested in this, over and over and over again? Much worse, that you keep hinting at the "owner" of this topic as being a potential spy with malicious intentions?

I am guessing you wouldn't like that if you ever opened a thread with something you may have offered to others. Having some dude implying you are not trustworthy (at best) and asking for an official alternative, over and over again?

If anything, I am trying to make you understand in a friendly way, not "moderate" you. But it looks pointless.

edit: posted before noticing the post right above

[Mike_Walsh]

Moderator note: off-topic&political post removed.

[Bilbo47]

Off-topic:

... you take on the role of moderator ...

When non-mod members step up to provide guidance, that should be a clue. The forum belongs to the community and not to mods ... thus community members have some obligation to maintain their interests here, which includes providing guidance about posts that bother community standards. Offloading mod workload is a good thing, especially when it's obvious that guidance is needed.

[Nuck-TH]

Point release is finally built. Autoupdate should be ready soon.

[BenFenner]

Thank you. The laptop serving the TV in our bedroom benefits from this release. \m/

[JayByrd]

I've noticed that the Slackware build on the Contributed Builds page has not been updated for a couple years. For myself, I've just been using khronosschoty's script as a starting point, and have been adapting it over the years to keep up with changes.

Since the last update to his SlackBuild script was for Pale Moon 33.2.5, optimized for SSE2, I "survived" the switch to AVX unscathed.

I have since acquired a laptop that supports AVX instructions and have started experimenting with new build parameters, and I have a couple of questions.

First things first: I want to follow official builds as closely as possible. From what I have gathered from the PM developer site for linux and the first post of this thread, I've arrived at the following parameters for three different builds:

Code: Select all

# For AVX (default)
ac_add_options --enable-optimize="-O2 -march=x86-64-v2 -mavx -w"

# For SSE2 (my daily driver, taken from khronosschoty's original SlackBuild script)
ac_add_options --enable-optimize="-O2 -msse2 -mfpmath=sse -w"

# For AVX2 (which I don't own, so have no way of testing)
ac_add_options --enable-optimize="-O2 -march=x86-64-v3 -w"

Question #1: Are these, in fact, the appropriate default settings for each?

Also, I noticed the following line in Nuck-TH's .mozconfig for linux:

Code: Select all

ac_add_options --disable-precompiled-startupcache

I've never used such an option in my personal SSE2 or AVX builds, so question #2 is: Is this option set for AVX2 only? (And if so, then is it strictly necessary, or just a preference?)

Thanks in advance,
JayByrd

[Nuck-TH]

33.6.1 builds are ready.

[Nuck-TH]

33.7 are finally up.

Not sure if changing key expiration date changed public key, but here you go if it does:

-----BEGIN PGP PUBLIC KEY BLOCK-----
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=Blyu
-----END PGP PUBLIC KEY BLOCK-----

[ron_1]

Not sure if changing key expiration date changed public key, but here you go if it does:

So I updated the .sig file, and then this is what I got. Is it correct? It says expired.

Code: Select all

gpg: Signature made Wed 09 Apr 2025 08:23:23 AM EDT
gpg:                using RSA key 5061CC51C94306050CFA1FEE48FAD2907D84EDEB
gpg: key 865E6C87C65285EC: no public key for trusted key - skipped
gpg: key 865E6C87C65285EC marked as ultimately trusted
gpg: checking the trustdb
gpg: public key of ultimately trusted key 865E6C87C65285EC not found
gpg: marginals needed: 3  completes needed: 1  trust model: pgp
gpg: depth: 0  valid:   2  signed:   0  trust: 0-, 0q, 0n, 0m, 0f, 2u
gpg: Good signature from "Nuck-TH <nuck*******@gmail.com>" [expired]
gpg: Note: This key has expired!
Primary key fingerprint: 5061 CC51 C943 0605 0CFA  1FEE 48FA D290 7D84 EDEB

[Veit Kannegieser]

Hello ron_1,
you want to select and save what Nuck-TH has hidden behind the [Show] button as file Nuck-TH.asc, and do
gpg --import Nuck-TH.asc.
It should report that it updated one key, with two new signatures or similar.

[ron_1]

I did

It should report that it updated one key, with two new signatures or similar.

Did as you said. Now do I change the name of the file back to palemoon_pubkey.gpg ?

[Moonchild]

Not sure if changing key expiration date changed public key, but here you go if it does:

I don't recommend changing the expiration date of keys. If a key is about to expire, you should generate a new key pair, because the published keys (in people's key rings, on keyservers, etc.) might not have the right data and not all systems accept updated keys once already expired. Yes, it's annoying that you would lose certifications/signatures on the key that expired, but you can use the old one to sign the new one to have an inherited chain of trust for people to rely on.
In general you have a choice to generate a key that doesn't expire (which would require you to revoke a key manually if compromised) or one that expires (requiring you to generate new key pairs). The change of expiry date is only an emergency workaround for some situations.