Jump to content

UCyborg

Platinum Sponsor
 View Profile  See their activity

  • Posts

    3,098

  • Joined

  • Last visited

  • Days Won

    28

  • Donations

    100.00 USD 

  • Country

    Slovenia

 Content Type 

Everything posted by UCyborg

  1. UCyborg
    Last time I opened my laptop, a cheap Lenovo G50-45 from 2014, I didn't see that battery and it also wasn't mentioned by a service manual I found at the time. So I wonder what keeps the clock going, the regular battery that powers up the laptop?
  2. UCyborg
    Revised CSS fix for Google Search: @-moz-document domain("www.google.com") { .lRPPDf { height: unset !important; } .VNzqVe { flex: unset !important; } } PixAI now deviates a bit from how it was intended, but at least it stops resizing like mad, though it would also need other fixes: @-moz-document domain("pixai.art") { #app .size-full { width: unset !important; height: unset !important; min-width: 100%; min-height: 100%; } }
  3. UCyborg
    I get this resizing crap ALL THE TIME with menu bar in with web app I have to support at work since many months ago, which have turned into years unless I unset a bunch of redundant CSS props. No one gives a s***. Try this with Google: @-moz-document domain("www.google.com") { .e6hL7d, .c8rBkc.oLJ4Uc, .xd0Ff.YWkDN, .RbVlwe, .I9e1O, .lRPPDf { height: unset !important; } } Perhaps a similar solution can be found for PixAI.
  4. UCyborg
    More observations, it does not take forever and it could've been like that since the overhaul. Started with bumping dom.max_script_run_time in about:config to 120 seconds and setting dom.always_stop_slow_scripts to false. I assumed 2 minutes will be enough for script to finish, the second setting was altered so browser doesn't stop the script, but presents the dialog asking user whether the problematic script should be stopped. So, official builds of Pale Moon at least come with PerformanceObserver disabled by default (dom.enable_performance_observer set to false). With that configuration, shortest time I recorded to load bank's login page was about 4 minutes, longest time was about 9 minutes and half. With PerformanceObserver enabled, shortest time was about 40 seconds, the longest was about a minute and half. The wonders of modern web development! That was all on a Raspberry Pi 5. Just another example of things going wacky on over-complex websites web apps.
  5. UCyborg
    The issue was first spotted 1 year ago. https://github.com/Alex313031/thorium-legacy/issues/4
  6. UCyborg
    Well, whatever the cause, the correct result is no crashing when used as-is with defaults. I noticed he has only one version of Thorium 122 for modern Windows (so no further patching of 122 from initial release), this one doesn't crash. Legacy one does, when run on Win10 or XP. Actually, it appears only Win32 flavor of legacy build crashes, Win64 legacy build is fine. And the one build for modern Windows is available only in Win64 flavor.
  7. UCyborg
    On a clean profile without any alterations and launching the browser without any command-line parameters/flags?
  8. UCyborg
    Actually, bank site works on another amd64 based computer, just doesn't work on a self-compiled build on a Raspberry Pi 5. I'll test later when I'm back home, have to try prolonging script timeout. I noticed the main script on that site has over 43000 lines of code (!!!) when you run it through https://unminify.com/.
  9. UCyborg
    Tried performing payment with Paypal, didn't notice anything unusual, went through as usual. Maybe it's only noticeable on really old computers? Can't enter my bank with Pale Moon anymore, the script hangs forever. Maybe I should start thinking about goin' somewhere else, first they made a retarded mobile app (clunky and bloated), at the same time they made retarded website which looks it was made only for mobile devices, they they cancelled the old mobile app and website and forced all users on new retarded versions. Afterwards, they obviously broke website further.
  10. UCyborg
    Yes... https://github.com/Alex313031/thorium-legacy/issues/62 The issue was closed, but it wasn't solved in legacy build.
  11. UCyborg
    Oh, I don't consider whatever bad happens with content user wants to post to be "an editor issue". What happens after submitting is no longer under the client control.
  12. UCyborg
    Recently read about this one, it appears to be one of the more interesting forks of Mozilla Firefox. More knobs in settings, sidebar, keyboard shortcuts; among other things, there's a knob for tab bar below toolbars, some update settings controllable through GUI like in old Firefox versions (there's no "Never check for updates", but the option to choose to install them is there), JPEG-XL support enabled by default, knob for turning on userChromeJS script support... There's also Web Apps feature, I guess equivalent to Chromium's "install website as app". Available x86 and x64 builds for Windows, x64 and aarch64 builds for Linux and a version for macOS (guess whatever's current in the Apple world). Homepage: https://floorp.app/en Downloads: https://floorp.app/en/download Installation instructions: https://docs.floorp.app/en/introduction/installation/
  13. UCyborg
    Cloudflare serves cookie consent script for the site, the rest are ads. I block all with extreme prejudice, the editor works fine, apart from the occasional "post goes through although it doesn't show until you refresh". Sometimes, but rarely, there's a delay when posting (server/database overload?).
  14. UCyborg
    It was my assumption platform development subforum is suitable if you can point out the direction where the issue lies. But issues not being acknowledged also happened in the past in web compat subforum. Though I think you have a point; the latter might slightly more suitable just due to the fact that subforum is all about real examples of broken sites. It's my hunch them being bombarded with issues all the time makes them indifferent at times. And, well, that's just how it is in open-source world.
  15. UCyborg
    You already did, it broke WebGL for you (and me). I get million errors in chrome://gpu, Google Street View is useless as nothing renders at all. Old dev build of Chromium 39 is fine, at least old WebGL games work and are playable, HexGL, QuakeJS etc. Forgot to try forcing OpenGL, doesn't work in old Chromium here and I guess it's a long shot with these hacks as well, but OpenGL is the only way to access full potential of DX 10.1+ GPUs on XP, at least in theory. There is improvement in CSS animations, eg. the ones on this site on people's profiles, where you can click the big picture and it expands, it was very choppy before, now they're smooth. YouTube is better, at least you can tell 720p@60FPS from 720p@30FPS, still some frame drops, maybe it's just due to overhead as XP lacks desktop composition. NVIDIA drivers for XP never inspired confidence in my experience, so who knows much they are or aren't a factor.
  16. UCyborg
    @IDA-RE-things XP x64, updates 'till about 2012 mostly, AMD Phenom II X4 920, NVIDIA GeForce GTX 750 Ti with driver version 355.98. Can't say I can decipher what your DLL is trying to do, it looks like chrome_elf.dll tries to LoadLibraryW(L"kernel32.dll"), but when I step into CALL DWORD PTR DS... instruction where OllyDbg mentions LoadLibraryW in the pane below disassembled code, I end up someplace else in one of your DLLs that just calls ExitProcess(0).
  17. UCyborg
    Yes, the browser doesn't load at all with it. Genius, zero CPU usage, deserves a Nobel Prize.
  18. UCyborg
    Hm, this might not be an actual valid option, seems you just omit/comment out the line with --enable-jemalloc in .mozconfig to achieve the effect. So far, I got a crash on one website when using an inspector, I think it was actually https://pixai.art/. It appears it's not an OS specific issue.
  19. UCyborg
    I know about that workaround, but it shouldn't be required. I can also confirm @NotHereToPlayGames's observation that it tends to occur with sites specifying font sizes in em rather than px. I noticed discussion a while back on GitHub about potential GDI issues, but I don't know if they got anywhere. Also, version 122.0.6261.168, after testing on different computers, I've came to the conclusion that both SSE2 and SSE3 versions actually crash everywhere for me on Google Drive, Win10, Win11, Win XP, doesn't matter, on multiple different computers. Tested with clean profile, no settings were changed, no extensions installed, just opened the browser and went straight to Google Drive. Hey, my motherboard's from 2008. I do wonder if updating XP x64 beyond the year 2012 would help with anything. Though that would take more effort than I'm willing to invest at this point since my XP refuses to update by usual means after installing Legacy Update. Even tried with an old backup of XP installation from few years back. Who knows what's wrong with it, 'been pretty careful with it as I am with other OS installations. The only other idea I have, I used to turn off advanced text services back in the day, this turns off ctfmon.exe. Seems like a long shot, but since both Thorium and Supermium mess with USB despite their actual USB functions being completely non-functional on XP...
  20. UCyborg
    Both Thorium and Supermium manage to destabilize my XP's USB stack to the point it's not detecting things being unplugged and plugged until the browser is closed. Thorium specifically is useless for Google Drive here as well, tab crashes as soon as I'm in. Both are buggy in fonts department (from disappearing fonts to incorrect rendering).
  21. UCyborg
    Chromium just tends to be the most popular for some reason. Security vulnerabilities are popping up all the time in it. Regarding functional issues, more people using it, more combinations of things they click, more whining about banal things etc.
  22. UCyborg
    There was a suggestion in this thread that freeing memory might be improved if one closes whole window. Got Pale Moon to consume little over 4 GB yesterday, having it open throughout the day and browsing as usual + a bit of GitHub, the site that is notoriously problematic on UXP browsers. In the end, I opened an empty window and closed the old window. The amount of freed memory after forcing it with Minimize memory usage button on about:memory page was negligible. my set of enabled extensions does cause a noticeable slowdown in browser startup time, but I don't recall them ever having other noticeable bad side effects. I got curious about --disable-jemalloc build option as it was brought up in one discussion on PM forum, so built the browser with it, will see how it works on Raspberry then. I know one Windows user reported a crash inspecting element on a certain site with such build that didn't occur otherwise. Maybe testing that way so it uses operating system's memory allocator would be an easier step towards finding some subtle bugs in the code. Apparently it was decided jemalloc was a better option back in XP days, might still be on XP, but the rest, who knows.
  23. UCyborg
    No, KeePass family of programs are about having a local database, KeePassXC has an option to add an additional factor to protect the database in form of hardware YubiKey, support for it was disabled as well in the new default minimal package. I know one online function has to do with retrieving websites' favicons, I don't know offhand if there's more functions that connect to the internet, but all these should be exclusively triggered by the user, not on its own.
  24. UCyborg
    Model tab looks closer to how it was intended. Except for uncontrolled element enlarging, I bet it's due to certain combination of CSS parameters.
  25. UCyborg
    Meanwhile in the world of open source... https://www.theregister.com/2024/05/22/apt_gains_keepassxc_loses/ TL;DR the controversial part, a software package under the name of keepassxc, providing binary distribution of well known password manager KeePassXC, was suddenly turned into a stripped down version by its maintainer and full version that everyone expected and uses was moved into a separate package. A whole set of features can be disabled using special compile-time flags. The package maintainer was rather undiplomatic about it, calling the full version crappy and development of included extra features misguided, citing security concerns while failing to provide evidence of any possible real-life security incidents. Some people got locked out of their password databases, others were surprised by suddenly ending up without the web browser integration etc. "Just read the NEWS, bruh."
×
×
  • Create New...