My Browser Builds (Part 5)

[kuja killer]

Hmm, i see. So that's another 2 weeks from now right ?? - Mathwiz/Roytam1

Edited June 8 by kuja killer

[Mehmed]

14 hours ago, UCyborg said:

Aw c'mon, another site all over the place due to the notorious new popular way of doing CSS, techradar.com.

https://www.techradar.com/phones/android/android-15-could-tell-you-how-long-your-phones-storage-will-survive-for

Also try https://www.politico.eu/ with a giant EU flag on full screen.

I am more and more convinced that UXP-browsers are absolutely not adapted for surfing the modern Internet. Considering that even primitive functions like proper SVG processing are beyond them.

Edited June 8 by Mehmed

[UCyborg]

Oh, so this is where Moonchild crew is going (regarding AVX builds)!

https://www.phoronix.com/news/Ubuntu-x86-64-v3-Experiment

https://www.phoronix.com/news/RedHat-RHEL10-x86-64-v3-Explore

BTW, H.265/HEVC video codec was brought up a while back, the only browser with working implementation that doesn't require GPU support for it I found is Thorium.

Edited June 8 by UCyborg

[UCyborg]

1 hour ago, Mehmed said:

I am more and more convinced that UXP-browsers are absolutely not adapted for surfing the modern Internet.

Modern web sucks with constant reliance on new bells and whistles.

Slightly off-topic, but I encountered another oddity with self-compiled build of Pale Moon on Raspberry Pi 5, enabling subtitles on https://piped.video/ makes the browser get stuck with full CPU core utilization.

Then go ahead and try being @roytam1 and roll your own builds.

[dmiranda]

8 hours ago, Mehmed said:

primitive functions like proper SVG processing

And to think I've been disabling SVG for decades now. Have I missed much, I wonder.

[UCyborg]

How do you do that?

[dmiranda]

gfx.font_rendering.opentype_svg.enabled - false

svg.disabled;false  false (sometimes default in many of ort browsers)

svg.context-properties.content.enabled;false (sometime default in many of our browsers)

svg.new-getBBox.enabled;false false(sometimes default in many of out browsers)

Can also play with cache size for svg, but I've found it unnecessary.

[UCyborg]

Unable to confirm any of these do anything to prevent SVGs from being rendered.

https://upload.wikimedia.org/wikipedia/commons/b/bd/Test.svg

No change on those two forums that use them that have them as part of HTML. Checked both official Pale Moon and Serpent 52.

[dmiranda]

Those settings used to work, on a tight, secured browser (I'm writting from mypal 68, with settings for social media) used to break sites as linkedin. I will give it a try with sp53 (which I still run very constrained) to see if I was using other settings, as well.

[dmiranda]

Yup, now stands as ||*.svg^$script,document,important in ublock, when using Tor, can't remember why the set up -I think to remember back in the day some potential hacking being done through svg.

[UCyborg]

OK, that would block standalone SVG files. I was curious how that like button on those forums would behave, eg. would the image be invisible, but still clickable.

BTW, an open issue for CSS layers: https://repo.palemoon.org/MoonchildProductions/UXP/issues/2486

There is a server-side polyfill for this feature: https://www.oddbird.net/2022/06/21/cascade-layers-polyfill/

But UXP browsers aren't listed in any compatibility table web developers are looking at.

[user57]

well someone opened the image format question over SVG

we made it to the heic image format (both software and hardware register based (not gpu (but guess what they are useally the same or the software controlled ones are useally a bit better because you can set them + being upgraded)), the same encoder can encode video too 

what we should realize that we cant skip the encode timing so easy

actually it makes a very big difference how complex (aka slower) you set the encoder

someone already did a graph showing that

https://user-images.githubusercontent.com/968944/195952757-cd1cdab4-6c8e-46a3-b5ed-de7fcce1371f.png

https://user-images.githubusercontent.com/968944/195952806-758d0240-020b-4936-ba09-d79e62bf0d3f.png

very easy to see it that the image quality (measured in DB (higher is bigger 42 + are useally very good) increases using slower settings

 

while the "fastest" is going with like less DB the "slow" is going with increasing DB

so i actually did not want to have the settings faster faster fast ect. 

because to me it seems a fault to set these 

 

but we have some advantaged not only slow we can set the best (placebo) (strukturag only use slow later on)

 

 

the increase in image quality increase for both (smaller file sizes) and (more image quality overall) using the slow/or aka the encoder settings that do more complex methods that aka take more time

 

so you can pass jpg in both better image quality and smaller file size

 

what i could do is using a pure RGB buffer (what is lossless) aka the BMP file format

 

but a raw file done to PNG is already lossless, im not certain about all compress settings for PNG but the big one (0) is lossless

 

so either having a .heic file you can convert it to a PNG file and see it lossless in the common windows image viewer, or making the .heic file to a jpg and also can open that file with the common windows image viewer

 

 

 

what i came to realize is that the heic encoder also can be done others encoders like the AOM encoder or the SVT encoder (those create a .avif file) 

the methods are very similiar, i think for video avif might have an adventage now because it can use more methods for video - i do not think that is the case for images - but if  someone can fix me up just do it

 

 

with a BMP file we would have a better editing method, we could make the heic file to a BMP file and overwork that BMP in the common windows editor (and then just compress it again) (+ a editor is exactly doing this (you just dont see it useally))

 

but having video encoder question actually it was actually more difficult to make an image then a video, i think we could do a video encoder also - but that opens a big question to control all the formats is a little to much 

but a simple one that supports 1-2 formats and not much of settings would be possible 

not to say that there are h.265 video encoders already out there

but the image encoder for .heic supporting xp was a new thing - and that engines removed up, no files needed at all (while the one from win10 need internal win10+ files)

 

https://msfn.org/board/topic/185879-winxp-hevcheifheic-image-encoderdecoder/

 

Edited June 10 by user57

[Slavich]

On 6/8/2024 at 2:28 AM, roytam1 said:

New build of Serpent/UXP for XP!NM28XP build:

Win32 https://o.rthost.win/palemoon/palemoon-28.10.7a1.win32-git-20240608-d849524bd-uxp-75b0995d95-xpmod.7z

I write through self-translation, so I apologize if I missed your English-language lines. Read more about the VK error here: https://repo.palemoon.org/MoonchildProductions/UXP/issues/2508

Thank you for your work for Windows XP users.

 

Edited June 12 by Slavich

[NotHereToPlayGames]

[roytam1]

New build of Serpent/UXP for XP!

Test binary:
Win32 https://o.rthost.win/basilisk/basilisk52-g4.8.win32-git-20240615-3219d2d-uxp-d835b252d7-xpmod.7z
Win64 https://o.rthost.win/basilisk/basilisk52-g4.8.win64-git-20240615-3219d2d-uxp-d835b252d7-xpmod.7z

source code that is comparable to my current working tree is available here: https://github.com/roytam1/UXP/commits/custom

IA32 Win32 https://o.rthost.win/basilisk/basilisk52-g4.8.win32-git-20240615-3219d2d-uxp-d835b252d7-xpmod-ia32.7z

source code that is comparable to my current working tree is available here: https://github.com/roytam1/UXP/commits/ia32

NM28XP build:
Win32 https://o.rthost.win/palemoon/palemoon-28.10.7a1.win32-git-20240615-d849524bd-uxp-d835b252d7-xpmod.7z
Win32 IA32 https://o.rthost.win/palemoon/palemoon-28.10.7a1.win32-git-20240615-d849524bd-uxp-d835b252d7-xpmod-ia32.7z
Win32 SSE https://o.rthost.win/palemoon/palemoon-28.10.7a1.win32-git-20240615-d849524bd-uxp-d835b252d7-xpmod-sse.7z
Win64 https://o.rthost.win/palemoon/palemoon-28.10.7a1.win64-git-20240615-d849524bd-uxp-d835b252d7-xpmod.7z

Official UXP changes picked since my last build:
- Issue #2524 - Implement smart granularity for canvas poisoning. (359a4a2069)
- Issue #2522 - Part 1: Alias small, large, and dynamic viewport units to base viewport (5c06335c5b)
- Issue #2522 - Part 2: Implement support for logical viewport units (2ba31f4e97)
- [network] Perform a case-insensitive match on special cookie prefixes. (bfb171c783)
- [network] Avoid accessing raw pointers in nsTransportEventSinkProxy. (bcdf7dc845)
- [uri] Fix issues with external protocol handlers. (a09accc204)
- [DOM] Clone video info properly. (a2417d573b)
- [toolkit] Close autocomplete popup when window is resized. (b4caa69a9c)
- [js gc] Suppress GC during JSObject::swap (40c9102d6d)
- [uri] Implement additional filename checks. (9c8bd37fe5)
- [layout] Initialize PerSpanData members mBaseline and mReflowInput. (0e05cb9630)

No official Pale-Moon changes picked since my last build.

No official Basilisk changes picked since my last build.

My changes picked since my last build:
- Revert "PR #2514 - Don't split up happy emoji sequences" (97db731ae7)
- Revert part of #2492 and replace it with mozilla one. (ff8673e87e)
- ported from mozilla: Bug 1819025 - Don't consider following join-controls when determining whether a space is collapsible. r=emilio (fad4426b7b) (854fac5c4c)
- nss: update nssckbi to 2.68 (b794ec0afd)
- layout: fix corrupted unicode in comments, no code change (36bc963804)

Update Notice:
- You may delete file named icudt*.dat inside program folder when updating from old releases.

* Notice: From now on, UXP rev will point to `custom` branch of my UXP repo instead of MCP UXP repo, while "official UXP changes" shows only `tracking` branch changes.