UCyborg

So EVR is seemingly restricted in rendering videos in limited 16-235 color range on XP. Or is there a workaround or a missing update?

Looks like a Chromium with typical limitations like the older variants. Guess it's good to pay the bills if your whatever-provider's site sucks so badly. No need for polyfills. Also why is only Google on the built-in search engine list? It did crash on first startup on my XP x64 (only tried 32-bit version), but afterwards it was OK. OS is not updated beyond 2011/2012, except an odd crypt32.dll update from 2015. Laggy transitions/animations on websites, eg. changing pages on this forum, laggy videos (also doesn't prevent screen turning off during playback) etc. So CPU struggles while GPU has nothing to do. I see people being hyped up about Win2K support and what not, but, why? There's a charm in small footprint of period correct applications. Why would you want that monstrosity there? 32-bit chrome.dll is almost two hundred f***in' megabytes! 64-bit is obviously already beyond that. I don't know, maybe I'm just too old to "dig" today's kids' crazy ideas. PS.: Not this crap with the fonts again!

Please re-download, I reverted one condition that checked for Fennec ID (the old Android version of Firefox) as I'm not sure about correctness and would have to investigate deeper if the alternative was correct (as the condition was triggered on desktop browsers and I'm not sure if the data it got would be equivalent). As the changes don't target any desktop browser or mail client by ID, it would be good to test and see to make sure are no any new oddities in any case, Pale Moon/New Moon, Basilisk/Serpent, InterLink/MailNews or Epyrus. Also, not sure the about the versioning scheme Borealis Navigator follows, are there versions below 0.9 in the wild? I put the same min/max versions I put when I modded PassIFox, which was tested on that application at the time. Minor details, but I just like min/max versions reflecting what's actually out there. Obviously maxversion is a bit special and an asterisk indicates (merely to the user that happens to read it, at least that's how it is most of the time in practice) support for the whole branch of versions. SeaMonkey is also a bit special, even if it should work in the current 2.53.* version, theoretically, it could break anytime as they clean up stuff under the hood often. Anyway, I'm not a pensioner nor a lottery winner and am a slow person in general, so these things are a huge time sink, meaning I probably won't modify this extension further at this point in time.

You can try my version. It no longer depends on Thunderbird application ID, all relevant changes are in js/vapi-background.js, Hopefully all remains well with it, a one line change - adding Epyrus' application ID in one expression in the mentioned file - would have been easier, but the other way should be more correct.

You can probably get rid of dxva2.dll as well. From the version numbers, evr.dll included here is older (6.0.6002.18005) than the one I got (6.0.6002.18084), the latter is stable in that regard. Wonder why was Vista's evr.dll mixed with 7's dxva2.dll. Is there anything to gain from extra functions it has in XP environment? The only useful obvious addition is DXVA-HD. However that plug-in software processor may look like.

Long-term, this becomes less relevant as applications evolve. You can't expect eg. Thunderbird 52 extension to be compatible with application that is not Thunderbird 52 and doesn't do things Thunderbird 52's way. athenian200 was clear in that matter. But sure, you're not the developer and you're free to not look into the compatibility with Epyrus. It's always the problem of the extension developer, but yeah, gorhill decided years ago XUL won't be his problem anymore. So I guess it's really not.

Nope, that's not how this game works. Epyrus != InterLink && Epyrus != MailNews Therefore, if the extension doesn't work on it as-is, it must be adapted. Extensions must be tailored to the application they extend, it's always been this way.

Nice, you can also set title bar's text color, both for active and inactive state, I really missed that. Few small additions in that area would be nice, like a glow effect, text alignment, ability to have blend color taken from system's accent color.

Never installed the former and the latter doesn't have it. Apparently evr.dll (Enhanced Video Renderer) was bundled with .NET Framework 3, dxva2.dll comes as dependency. It can still provide some support functions to evr.dll on XP. EVR is selectable and functional in some players, but it was primarily made for Vista. Overall, things still look better in the land of supported to me. Seems the old version simply doesn't recognize that ID/string. It seems it didn't have the 4.6 requirement, still targeted 2.0 in addition to 4.0 according to .config file. The latest one craps out on non-existent method in .NET if you try that tweak.

This looks better: Driver version 355.98. This explains why LAV Video Decoder doesn't work on h.264. LAV has DXVA2 according to its settings, h.264 apparently works through DXVA1 on XP. Honestly, I've no idea how I have dxva2.dll in system32/SysWOW64, DirectX Redist seems a likely candidate, these files don't have normal custom file permissions, but inherited from their folders, no copy of them in dllcache folder either. DXVA2 is not supposed to work on XP according to Wikipedia, whatever goes through should only go through software, non-accelerated decoding paths at best. While dxva2.dll is loadable, when it comes to DirectX in general, you still need support from the layers at lower levels. Though it's interesting VC1 codec is shown for DXVA2 at all, I don't remember ever dealing with this specific codec, at least not knowingly. So output from GPU-Z makes more sense now, querying only DXVA2. That one VC1 variant in DXVA Checker also disappears for DXVA2 when renaming dxva2.dll to something else.

That's the one. I checked it when I was still running Windows 8.1. I'm not concerned about what GPU-Z shows on XP regarding DXVA.

Oh, I forgot about this one, but the other one I had in mind was focused specifically on DXVA and was a bit verbose about it. This is what GPU-Z shows here on XP x64: So I'd just have to wait about half-year and I'd have that. Oh well, I couldn't wait to play Wolfenstein: The New Order back then.

Do note that any code that runs continuously without pause on the CPU core will make the utilization on it show as 100%. The question is if the algorithm can be improved to complete sooner. The problem with single-core CPUs (which AstroSkipper uses as far as I'm aware) is that you effectively can't properly multitask on them, you can't give it work and do something else in the meantime. If you do, it will be interrupted for a short while to try to do your thing then get back to the other thing, but your thing will suffer immensely. Maybe less if you do something in the other program and give it a higher priority.

Doesn't this only apply to HEVC? I do remember there's a nice program out there to list exactly what works through DXVA on computer where it's launched, but I don't recall its name ATM. Yeah, I rarely watch anything new anymore. I used to worry about what my graphics card supports, if I waited one year instead of buying Radeon 4890, I would've got a card with DirectX 11. If I waited one year instead of buying GTX 750 Ti, I would've got a card that does video decoding beyond old H.264 and DirectX 12. I'm guilty of this as well (just the paragraph before), but Direct3D and DirectX are often used interchangeably. I didn't think of the X this way, but I knew that's how Xbox got its name - DirectX box. This goes way beyond Microsoft. I know because I work at the smaller company that does specific hardware and software. So practically, as far as I'm concerned, everyone does this.

OK, madVR doesn't work, but Enhanced Video Renderer (both variants) does. Can't use DXVA through either old XP compatible LAV Filters or new ones, PotPlayer can with its auto-selected Video Renderer, but it's very a bit primitive, put player on another screen and no picture. Nothing wrong with the player, just the platform limitation. CUVID works through through either LAV Filters (the only option with MPC-HC) or PotPlayer's built-in decoder. Whatever new codecs are supported with new LAV Filters, they obviously can't be HW decoded on XP era hardware/graphics drivers. My GPU is before the time they came with VP9 decoding, but I wondered if that is supported on XP with slightly newer GPU in any shape or form.

WhoTracks.me does not appear that heavy by itself, but the script putting links in the bubble takes a long time, 20 to 30 seconds for Google Tag Manager here, depending on the browser. The script is not large, but its logic takes time. You know, I can't help but wonder, if I buy a computer for 4000€, how much would it reduce the time it takes.

GreaseMonkey for Pale Moon 3.31.4 should be better suited for UXP browsers. https://github.com/janekptacijarabaci/greasemonkey/releases/tag/3.31.4Fork

I meant to say any new games, though there probably are and I'm just throwing out negative assumptions again. They're social butterflies I guess, something our kind will never "get". I've had bits of social life in the past, but it's all gone now. I also get sick of my own company at times, but can't escape from myself.

I bought a new car with the money made at work, though that purchase was mostly out of necessity to get by in day-to-day life. I doubt I'll replace the computer soon, I'm a bit tired of computers (using it at work), gaming doesn't feel the way it used to anymore and I love being lazy in free time more than ever. I'm dead after work and weekends are too short. Gaming would maybe be the only reason to replace the old computer, but I have too much old games and I'm not sure there's anything new out there for me anyway. Hm, https://www.verywellmind.com/asocial-vs-antisocial-differences-7555163

Found what breaks https://accounts.hcaptcha.com/demo for me. I have javascript.options.wasm_baselinejit enabled, guess that's still problematic on 32-bit builds. Looks like I was dribbling with hCAPTCHA before: https://forum.palemoon.org/viewtopic.php?f=61&t=24663 Back then they did something on hCAPTCHA side to fix it, although it seems this is general issue in the platform.

Yes, might have confused space with line break due to how Resource Hacker displays it... No. It shouldn't matter for newer systems either, it's just a convention since newer systems will pick the highest available entry, so you can order it that way to visually indicate what entry has higher priority. https://learn.microsoft.com/en-us/windows/compatibility/application-executable-manifest I haven't found the documentation whether 8.1 and 10 entries change anything in practice

Just took a better look at the files of this MPC-HC 2.1.2.18, looks like a KernelEx style hackery, not a pure Windows XP compatible port. I tried it on my usual XP x64 install anyway, but it doesn't appear to be usable here. There's no picture, just sound, the part where picture is supposed to be, nothing renders there, if I restore its minimized window, it's see-through with whatever was on that part of the screen at the time of window restoration. Same if I pick madVR renderer, which is surely functional as my usual PotPlayer can use it.

The first one was the most limited, the one from ungap should be the middle ground, the most complete is the one from core-js. Do the read the notes as it is said structuredClone cannot be polyfilled in its entirety. It's probably common for such code to be packaged in a way to be easily reusable in an environment not limited to a web browser and more easily maitainable/studiable by separating it into multiple smaller modules. The repo from ungap has it packaged in two ways (in order for it to be split in multiple modules), the one in cjs folder using CommonJS format usable outside of web browsers and the one in ecs folder using ECMAScript 6 standard's module imports using import/export in a web browser. Except I haven't heard of user scripts that are split into multiple modules, so going off from /ecs/index.js, I've put everything into one file, removing import/export and added the statement attaching the structuredClone function to the global object.

Right, there was a link posted earlier about that code, it's a special one, technically the kernel didn't crash, but it aborted execution since it detected user-mode csrss.exe process crash. Presumably it's logical the kernel wouldn't dump its own memory then as if one was going to go debug something in this case, it should be targeted at the crashing process as the kernel could normally continue if that other process didn't go down.

Forgot I already posted, there was no feedback from the forum software after clicking the button. Probably still goes through most of the time.