UCyborg
Content Type
Profiles
Forums
Events
Posts posted by UCyborg
-
-
On 11/14/2023 at 12:49 AM, dmiranda said:
There is a new version/fork that seems to do a better job: https://greasyfork.org/en/scripts/431573-youtube-cpu-tamer-by-animationframe
Has the exact opposite effect here, CPU usage is consistently higher with the script enabled.
2 -
Quotes shouldn't break anything in this case, though I guess generally in software they're strictly needed if to-be-quoted string has (or could have) (white)space characters as those could be interpreted as separators of whatever arguments you're passing. In this case, I think it can be safely assumed they won't come up with a feature name with a space in it.
Chromium doesn't use quotes when passing certain arguments similar in syntax like --enable-features among its processes, eg. --field-trial-handle=1956,i,12723921000856444030,14158919512122076707,262144. In other cases where quotes are needed, it seems to quote the entire argument, eg. "--database=C:\Users\User\AppData\Local\Chromium\User Data\Crashpad". I think it's not something to make a fuss about.
0 -
11 minutes ago, VistaLover said:
FWIW, you can always use a new St52 profile with just uBO (or other script blocker), specifically for Discourse-based forums (in any case, you can at least try this to verify my solution really works
) ...
BTW, the user agent override looks redundant.
1 -
I reproduced http://www.quakejs.com/ issue on Windows 10 VM in both Supermium and Ungoogled Chromium, it has to do with my inconsistent DNS config across different OS installs as it doesn't happen on my main Win10 install, I often connect through the phone connected to home network wirelessly and there I have phone's IP (the one exposed when USB tethering is enabled) entered manually as DNS rather than target DNS directly. VM was using AdGuard DNS IPs directly.
No DNS over HTTPS here, the explanation is on below link, some DNS record was forcing redirect to HTTPS, but that site is only HTTP and trying HTTPS on it redirects back to HTTP:
https://stackoverflow.com/questions/73230856/307-internal-redirect-non-authoritative-reason-dns
The current flag to turn off the feature in Chromium to look for those records to redirect to HTTPS is at chrome://flags/#use-dns-https-svcb-alpn. AdGuard's DNS server must have that record for this specific website.
BTW, no issue with font decoding in Supermium running on Windows 10.
12 hours ago, NotHereToPlayGames said:So that US Flag could just be indicating that the web site being connected to is located in the US. As opposed to indicating a VPN using a US IP.
That is correct, the flag comes from IP Domain Country Flag extension.
0 -
10 hours ago, Dixel said:
"Looking ahead: Manifest v3"
"In 2019 we will introduce the next extensions manifest version. Manifest v3 will entail additional platform changes..."
I think it was already discussed at some point, isn't this browser just one or two versions behind (Chromium version) the one that started supporting Manifest V3 extensions?
0 -
Maybe I exaggerated with barely works, most of the browser probably does work OK, I'm just curious about the fun stuff. It was Supermium 118.0.5991.0. There could be an extra step needed that was maybe buried in some thread on the forum since nobody mentioned the font issue here, but I basically started with Vista SP2 patched with all up-to EOL updates, then added KB4474419 cumulative update, the one for Server 2008, then vistaexkernsetup_03092023.7z. With that setup, it complained about mfplat.dll in chrome://gpu, so installed mfplatsetup.7z as well, just in case.
0 -
This browser doesn't even know about MV3 extensions so they (MV2 ones) should continue to work I think. Presumably the older ones may continue to be available on https://www.crx4chrome.com/, at least those that are there now.
But I have a feeling y'all forget about old MV2 extensions the same way you've forgotten about 360Chrome 11.
0 -
XP? It barely runs on Vista. Downloadable fonts don't seem to work here, I get errors about browser being unable to decode them. Counter-Strike's UI looks really funny, letters or squares instead of icons/images. Also observable on this very forum.
That and the extended kernel broke 360Chrome 13.5, GPU acceleration is non-functional there (GPU process crashes) and almost all extension processes crash on startup.
Though the best Supermium manages is WebGL 1 (2 is non-functional) through ANGLE, only D3D9 (like 360Chrome without extended kernel), native OpenGL can be forced, but slow due to some DXGI function failure, so practically useless. Haven't experimented with NVIDIA drivers beyond the official 368.81.
Vista's D3D11 always seemed more bare-bones to me compared to more feature complete implementation in Windows 7+.
Edit: Right, forgot there was trouble installing extensions first time around, some error to do with trust, didn't note it down...I guess the kernel extension crapped out, was trying out COD:MW Remastered before (without DRM, normally runs on Win7+), the browser started accepting extensions after reboot. But the strangest problem I've encountered, http://www.quakejs.com/ doesn't load, ERR_TOO_MANY_REDIRECTS, even with clean profile.
0 -
9 hours ago, NotHereToPlayGames said:
His reference is to "DiscardVirtualMemory".
A minor setback in my view, but technically/programmatically a "h@lf-a$$ed" scheme that literally just "ignores" arguments assigned to the function yet those arguments remain in the code.This topic drifted off discussing Supermium, which I haven't mentioned specifically, but I didn't deraiI the topic, just joined in.
But folks here like to throw in other Chromiums into the mix.
Anyway, I was referring to the one hosted on http://xpchrome.com/, only later https://github.com/weolar/xpchrome was brought into the mix. The one from first link certainly can't launch on Windows 7 and earlier since DiscardVirtualMemory is referenced directly in chrome.dll's import table. Not sure about Windows 8.1, is there an update that adds DiscardVirtualMemory? I only ever run Windows 8.1 that didn't have it, though MS documentation says it's supposed to have it, I only ever saw it in Windows 10 and comments in Chromium's source code say the function is buggy in older builds of Windows 10.
But actual XP compatible Chromium 115 on the second link, that's new. I wonder if it keeps screen from going off while watching videos or using Screen Wake Lock API. Older builds of 360Chrome 13.5 couldn't do that (on XP), not sure if anyone paid attention to that in newest build.
0 -
2 hours ago, we3fan said:
Hi UCyborg, is this the program you use? (PEview 0.9.9)
No, but peview.exe bundled with System Informer/Process Hacker.
1 -
7 hours ago, AstroSkipper said:
All my user interface modifications can't be done by using web extensions due to their damn restrictions in the more recent versions of Firefox.
I don't see it that way, it says in their name, "web" extensions. You can't expect "web" extension to change the "browser" since they naturally target mostly (with some exceptions) web pages.
0 -
@seven4ever
I can't say I notice a difference between Linux and Windows when it comes to general surfing these days, so if I just look at load times. I do remember about 10 years ago on this 2009 desktop, which had an older GPU back then (which drivers sucked for gaming on Linux), Firefox was noticeably more fluent on Linux than on Windows 7. Web was much simpler back then. Things start to differ these days when you start looking at WebGL/video decoding and such, very dependent on GPU drivers.I'd still go Linux route if I had the problem with newer Windows rather than mess with h@lf-a$$ed Chromium backports. You even have working native Widevine on Linux these days for DRM streams, which wasn't the case years ago. While I can live without Widevine, I'm the only one in my family who can.
But I also don't have any attachment to legacy Windows either.
6 hours ago, woi said:Just looking at chrome.dll in PE Viewer, no effort was done even for Windows 7 compatibility, imports DiscardVirtualMemory like official Chromium. So, nothing to see here.
0 -
You know more tricks than I do. It's weird, how come Google Drive is broken (loads, but can't open files/folders, context menus etc.) if I have all filters' checkboxes off, but works with BYPASS FILTERS button on.
Though I'm just a simple user and can't imagine myself ever being able to make extended use of Proxomitron.
0 -
Guess it's sort of similar how some devs don't do x86 builds anymore, Microsoft also doesn't bother with x86 version of Windows 11 and since roytam1 targets old systems and there isn't much difference in practice when using email client in x86 or x64 flavor and since he builds other browsers/mail programs, he might have just opted to provide 32-bit build only since it'll work on both.
I'm still a hybrid, use some programs in 32-bit flavor and some in 64-bit flavor. There can be differences in favor of x86, older CPUs may be quirky. I find Pale Moon to be a bit more responsive in 32-bit flavor on my system. Somewhere it was mentioned less code will fit in CPU caches since x64 instructions are generally larger, so this could make a speed difference in certain scenarios in favor of x86.
Though I never went into deeper comparisons, I do remember several years ago, when I was experimenting with DarkPlaces engine that runs the old Quake I and other games, 32-bit build produced slightly higher frame-rate while 64-bit build loaded map and all resources a bit faster. I was using high-resolution textures obviously, can't make good comparison with the load for computers from the 90s. But differences were small, load time 2 seconds at most and with FPS, also a frame or two if I recall correctly. Though I don't remember the frame rates anymore...frame times are more interesting since difference between higher frame rates is much smaller than difference between lower frame rates.
On x64 Windows, there's always some extra work in the background for 32-bit applications, lookup WOW64. But when the code starts flowing and CPU is switched to 32-bit mode, it's as fast as CPU manages, until some system call/context switch.
1 -
Nice, that URL would have been good to know back then. Shorts should still be in English in 360Chrome, right, eg. ERR_SSL_PROTOCOL?
Before I forget, I got the extension for decoding JPEG XL images working some time ago. Accept HTTP header may not be sent as intended if there are other extensions that mess with Accept HTTP header because old Chrome versions don't allow appending to that header, can only replace it. Appending is supported since some version after 100 and before 109, I don't recall the exact version, there's an issue on their bugtracker about the subject, again, can't find it ATM. May not be that much of an issue in practice since it seems JPEG XL support is preferred to be checked through JavaScript. Also, the decoding library is implemented using WebAssembly, so you don't want it disabled for it to work at all.
BTW, I was messing with Proxomitron a bit and got translation feature working in Edge 94 by just changing headers advertising browser version[*]. It's also blocked in Edge 94, server edge.microsoft.com just returns error 400 (bad request). Maybe the same approach would work for old Chrome 86 on whatever server it connects to to translate web pages.
I wouldn't recommend using Proxomitron in general because things on some sites break just because the browser communicates through proxy, even when no changes are being made. Maybe it would be possible to put together a web extension to do the job. Could it be enough just having base manifest.json with basic extension metadata and info for targeting right URLs coupled with rules.json to do the replacements in HTTP headers?
Or perhaps looking into ModHeader extension, which is specifically for dealing with HTTP headers. I suspect digging out an older compatible version will be required.
Edit: [*]For Edge, it's enough to just change User-Agent header sent when GETting edge.microsoft.com/translate/auth, though there are also two other headers advertising browser version.
Edit 2: Right, Edge uses different translation service. It's not so simple with Google Translate.
0 -
Very late to the party, should've done it way sooner, but finally got myself extra 2 GB of RAM so I'm at 6 GB now and immediately feeling less constrained. DDR2, ancient tech, difficult to get, especially if you don't fancy buying directly from strangers.
Computer turned on normally after installing it and off to Memtest86 5.01 I went, which was installed alongside Kubuntu 21.10. As soon as it started testing 4 GB - 6 GB range, endless error spam! Then I wasted time messing with BIOS settings such as timings and voltages, no improvement. The new-old stick is from another brand and has slightly worse timings, so the old sticks have to work a bit slower. Then I swap them around, put the original 2 identical sticks in the two slots most further away from the CPU and the new stick in the very 1st slot near the CPU, all related settings in BIOS to Auto, same problem, again only above 4 GB, where memory space of one of the old certainly working sticks reside now. What the hell, that makes no sense.
Then I take multiboot USB flash drive prepared with YUMI, which also has Memtest86 5.01 on it. With this one, it magically works, entire pass done with zero errors. Back to Memtest86 installed on the hard drive, errors in 4 - 6 GB range immediately. I repeat the tests again with original layout, so the old sticks in the first two slots and the new-old one in the 3rd slot with same results.
Now I'm really not sure anymore what computer issue I encountered in the past is the most bizarre. Just wanted to share this experience in case anyone encounters something similar. The RAM works fine. If there is a technical explanation, it would be nice to know it, but otherwise, if someone encounters this, a different installation of Memtest86 is worth a shot.
Maybe it has something to do with GRUB, which is how Memtest86 on the hard disk was launched while the one on USB flash drive was launched by SYSLINUX. But Kubuntu launched by the same GRUB certainly works without errors, I ran memtester (user-mode memory tester program) for a bit, got it to allocate 5 GB of RAM, which should definitely cover some portion of the extra stick.
Another person experienced the same issue with a bit more capable hardware, also AMD based, but AM3 CPU socket and DDR3 RAM and also using Memtest86 from Ubuntu, though he runs it from its install media, so either DVD or USB. One motherboard has AMD chipset, other has NVIDIA. Perhaps something to check if it still occurs with latest and greatest and check other distros that come with Memtest86. I wonder if it's a combination of multiple factors.
https://forums.tomshardware.com/threads/memtest86-errors-only-above-4gb.3751816/
0 -
Maybe my computers are too new or my expectations for Linux are too high, but I never understood where it got the reputation that it's good for old hardware. Things just look poorly optimized on Linux from my POV. Have a laptop from 2014 with a poor dual-core 1,35 GHz APU and damn, could scrolling in web browsers feel any heavier? Don't get me started on video in a web browser, LOL. Though TBH, I haven't experimented with lightweight desktops, but then it's 2023 and don't see a reason why should I settle for anything less than KDE. But I'm not sure that would help with anything but RAM consumption and from what I've read some time ago, KDE at some point even surpassed certain "lightweight" desktops in performance department. I don't consider RAM consumption an issue, could get 8 GB of RAM easily if I wanted, currently it only has 2 GB, upgrading just wasn't a priority due to just occasional lighter use without opening too much stuff at once.
0 -
YouTube CPU Tamer had the bug with spinning circle during playback ever since I first tried it. Don't remember it ever having any significant effect on CPU usage. The other YouTube CPU Tamer by AnimationFrame script seems to have the exact opposite effect, at least on Pale Moon.
0 -
I like Interlink's stock UI the most. though I could do with Epyrus with ClassicTB2 theme.
I just checked the latest version of Interlink, which I think is 52.9.8194. OAuth2 works with GMail, maybe it was another mail provider that works with roytam1's fork, but not official version. That makes Epyrus not working with it out-of-the-box even stranger. I must have read about another mail provider that Tobin couldn't get something for it to make Interlink work with it.
0 -
The name of that project really doesn't say anything to those that don't know it, OneCore is a library that Windows application developers may use to link their applications against.
11 hours ago, NotHereToPlayGames said:CAN
I know this one, Controller Area Network.
The previous car of mine must have used K-Line.
1 -
On 11/5/2023 at 11:59 PM, Jody Thornton said:
Still I prefer to use an application natively, rather than through x86 emulation if I can. I don't see that as "just because". However, even if that were the case, I still prefer to run an application that utilizes the native bus-width of the OS I'm using. My choice - good enough reason for me
Indeed, though I find it curious that lack of 64-bit binaries bothers you, but not catering down to the lowest common denominator on the API level, which is the main thing with roytam1's programs.
Anyway, my suggestion would be Epyrus or SeaMonkey. They both come in 64-bit flavors and don't cater to ancient operating system versions. SeaMonkey is a full internet suite, so comes with web browser as well, though some are concerned about its future due to very small dev team and how they're diverging from Mozilla's platform. Epyrus would be to Pale Moon like what Thunderbird is to Firefox, it's developed by one of the people among Moonchild Productions crew. They say underlying platform code updates shouldn't play the big role when it comes to mail client (at least for now I think), so the core stuff should work OK even in older versions of whatever fork you end up using.
There's one specific about Epyrus, it doesn't pretend to be Thunderbird for legal reasons like Mozilla forks tend to, meaning if you need it to work with GMail, you'd have to go through the procedure of generating your own OAuth2 client key and secret. I didn't finish it through to the end and ended up deleting everything I created, wanted to keep things simple, Google can think whatever they want... Open Config Editor and set the following prefs:
oauth2.google.clientid -> 406964657835-aq8lmia8j95dhl1a2bvharmfk3t1hgqj.apps.googleusercontent.com
oauth2.google.clientsecret -> kSmqreRr0qwBWJgbf5Y-PjSUThey aren't really a secret.
0 -
Version numbers don't say much in roytam1's world, at least in the following 3 browsers, it's the date that's more important. Upstream Basilisk directly uses date for version number, but for these forks, from my understanding:
- Serpent 52.9.0 - browser like Firefox 52.9.0 ESR + updates by the MCP
- New Moon 28 - browser like "new" Pale Moon
- New Moon 27 - browser like "old" Pale Moon
BTW, I noticed actual Firefox 52.9.0 has more animations in GUI, like when you click the downloads button, popup animates into place. Similar the popup of uBlock Origin extension when you click its button. Is this enablable in any descendent browser?
7 hours ago, adata said:This site tries to, among other things, create over 16 WebGL contexts! Intentionally or due to its code crapping out?
5 hours ago, NotHereToPlayGames said:When it comes to browsers, it is VERY COMMON for OLDER browsers to be faster.
While doing less than new browsers, as you pointed out.
14 hours ago, Mathwiz said:I wonder if a FF 52.9 user agent is all that's needed to dissuade Facebook from sending all the Googlescript that slows UXP down so badly?
Haven't tried the user agent on newer browser, but actual Firefox 52.9.0 doesn't seem to work with Facebook.
4 -
8 hours ago, j7n said:
I missed the meaning of upstream and was looking at New Moon and Serpent.
I was asking if it exists in New Moon or Serpent. I only checked thoroughly when I edited my post above, that's when I noticed commit roytam1 pointed out above and that's in fact not available in roytam1's New Moon.
If I read the code right, it shows after 13th October and before 5th November. I initially assumed it was during 31st October only.
0 -
2 hours ago, j7n said:
How long to wait?
Just few seconds.
Edit: Nope. Looks like part of official branding.
0
Changing SK_GAMMA_CONTRAST without recompiling Chromium
in Web Browsers
Posted · Edited by UCyborg
Over 90 occurrences on each scan, one searching for float representation of 0.5, one for double representation. Looking for a needle in a haystack.
Windows version doesn't matter. Yes, they are thin on Vista and you not seeing it doesn't change that fact. We might interpretate "thin" differently, but either way, they're also like that In Firefox based browsers when explicitly selecting Skia backend, which is fortunately not default. Though the alternatives for Direct2D backend on Windows versions before 7 still suck in my eyes, Not a problem for me since I'm on 10 anyway and downgrading to anything older than that would bring more problems than it would solve.
I use Chrome Super Font Enhancer on non-Edge Chromium browsers, the picture on that page demonstrates the difference. The default font rendering hurts my eyes with prolonged reading.