I noticed that DNT is being sent by default:

I don't really prefer having such a header enabled by default (it should be left to the user IMHO), but if you want to send by default, you should use its successor GPC instead of DNT which is pretty much obsolete and useless.

The relevant pref is privacy.globalprivacycontrol.enabled

I was checking mercury's benchmark data to see how it fares against thorium but it seems the data mentioned on the website here is exactly the same as mentioned in Thorium, maybe there is some mistake that i believe woud want to be addressed

https://www.the-independent.com
Videos "powered by pixels.ai" won't play. They don't in Troubleshoot Mode either. Other sites work properly with Mercury, other browsers work fine for the-independent.com

The browser has it's own unique user agent, you can see that by going to https://ifconfig.me/all
The problem with this is that:

1. It leaves a unique fingerprint
2. It breaks sites like Google that do agent tailoring

I want to make Mercury Portable.
What should I do?

Installed 115.4.0 after I saw Chris Titus's video on Thorium. Norton & Malwarebytes went nuts during and after the install calling out (Norton) accessing protected files and (Malwarebytes) phishing and malware being blocked. Was not thrilled about that and uninstalled it.

Mercury feels quicker and takes less memory than official Firefox. But 110.0a1 occasionally crashes silently for me.

Wanna update to latest version but start from 111 extensions.unifiedExtensions.enabled no longer works, leaving the obstinate Unified Extensions button along. Just wonder if you have any plan to add this setting back in the future.

Thanks for any reply.

Hello, i hope to use this extension to open media player to play youtube videos and video from other sites :

https://github.com/andy-portmen/external-application-button/

the problem with firefox based browser, we can open video of the browser wih media player but there is child process in the background and when we close firefox, the media player closes , we hope our beloved Mercury browser will fix this big problem, chromium browsers doesnt have this problem, thank you !

Thank you for developing Thorium and Mercury.

Would you consider maintaining a macOS build for Mercury as well?

30 seconds after installation, windows defender deleted the .exe without user input, flagging it as a "severe" level threat.
I believe that it's not malware, but i'd rather not turn defender off.

Trying to build the browser fails with a vauge error.

....to this resources in background:
content-signature-2.cdn.mozilla.net
detectportal.firefox.com
firefox.settings.services.mozilla.com
push.services.mozilla.com
location.services.mozilla.com
contile.services.mozilla.com
aus5.mozilla.org
ciscobinary.openh264.org
and so on

Consider the following image: https://people.csail.mit.edu/ericchan/hdr/jxl_images/20140606_102418_IMGP0297.jxl

This image is an HDR image tagged with BT.2100 primaries and PQ transfer. Mercury appears to be requesting BT.2100/PQ pixel data from libjxl, and then subsequently rendering that pixel data as though it were sRGB pixel data, which is incorrect. My monitor is SDR/sRGB.

Possible solutions:
(1) Request BT.2100/PQ and then send it through the color management engine as BT.2100/PQ
(2) Request sRGB from libjxl anyway, if sRGB needs to be the final target.

Somewhere along the line, HDR pixel data is being requested from libjxl and then it's being treated as though it were sRGB. You can do this explicitly by decoding the image with djxl via:

djxl 20140606_102418_IMGP0297.jxl --color_space=RGB_D65_202_Rel_PeQ djxl.png

and then use a utility like tweakpng or umbrielpng to strip the cICP chunk and replace it with an sRGB chunk.

Then, if you view the sRGB-tagged image in an image viewer, you can see that it looks exactly like the JXL file does when displayed via mercury, which means this is the problem that's occurring in the render pipeline somewhere.

I don't have a password on my Windows account, only pin (for better security), and Firefox/Mercury .exe installer shows this window:

There is nothing I can do, since I cannot log into anywhere without a password. But Firefox .msi installer does not have this issue, so I use that. But there is no .msi installer for Mercury. Perhaps it would be nice to have.

Extensions cannot be installed

I installed Mercury from the main repo. After launching it, I didn't have any issue immediately. After five minutes of using it, my antivirus started to flag it as a trojan:

Event: Object deleted
Application: Mercury
User: -----
User type: Initiator
Component: System Watcher
Result description: Deleted
Type: Trojan
Name: PDM:Trojan.Win32.Generic
Threat level: High
Object type: Process
Object path: C:\Program Files\Mercury
Object name: mercury.exe
MD5: -----

This has happened at least 4 times.

Why did it happen? Should I just add an expcecion to my antivirus or is there something that's not working properly with Mercury?

The first one should be relatively easy, simply removing the check, but I don't know how involving would be making them work on extension generated pages, though that would be important improvement for usability

Nixpkgs package manager & build system enables static linking for almost everything, while (as I know) generally more features work compared to Snap/Flatpak. Nixpkg application can work on any distro for a given architecture. Same goes for Thorium, I would like to see a nixpkg for it as well.
I would also like to see SSE3 releases for Mercury.

Detected as Trojan:Win32/Bearfoos.A!ml in zip, so I've check VT hash and found
Installer detect: 12/68
Zip detect: 10/60

First of all, I think I've downloaded 115 and extract overwrite, but in about dialog say I'm still use 112, so I clean entire program folder, extract and open browser, then got alarmed.

System: Windows 10.
Left - Librewolf, Right - Mercury

P.S. Have issues with firefox sync. Am i only one?

First off, love your browser forks! Keep up the fantastic work!

The Windows builds are missing from assets in this release

Is this feasible, or only possible if you rebase Mercury off Firefox Beta/Nightly?

The way I understand it, Mercury currently targets x86-64-v2 + AVX. It would be cool to have a version targeting the full v3 list, which includes among other things AVX2.

https://github.com/Alex313031/Mercury/releases/tag/v.115.0.0
Version tags should be the same as the bin names, not doing this makes packaging break
Example:
https://github.com/Alex313031/Mercury/releases/download/v.115.0.0/mercury-115.0.en-US.win64.zip
Should be:
https://github.com/Alex313031/Mercury/releases/download/v.115.0.0/mercury-115.0.0.en-US.win64.zip

Mercury don't recognize windows theme & don't change titlebar accent color with dark theme.
First image - Firefox, second - Mercury.

I discover Mercury browser with pleasure !

but I don't see a link to download it ?

I hope that Mr Alex will make changes to reduce the data used by the session store and session restores, there are a lot of parameters relating to the session store parameter that I no longer know for sure which parameter to disable unfortunately :(
I also hope there will be an option to disable MSE media source extension quickly so we can buffer full video in youtube and some streaming site, pale moon browser has option to disable media source extension, and may be Mercury can be inspired by Basilisk browser which has recently been updated and the project relaunched:

https://www.basilisk-browser.org/

or the amazing Floorp browser make by japeneese students community, this is new updates adress !

https://floorp.app/
https://github.com/Floorp-Projects/Floorp

and ice raven browser :

https://github.com/fork-maintainers/iceraven-browser
and also to enable quickly D3D9 beacause i because I tested this setting on chromium and the browser becomes fast, as well as an option to disable javascript by default to avoid malware, and option for adjusting performance settings to reduce memory can be done in Preferences or by changing the dom.ipc .processCount value to 1-8 and browser.preferences.defaultPerformanceSettings.enabled to false manually in about:config.

However you may want to manually adjust this setting to increase performance even further or decrease memory usage on low-end devices.

https://wiki.archlinux.org/title/Firefox/Tweaks#Change_Performance_settings

and also disbale prefetch and captive portal !

thank you very much !

and Ghostery
https://github.com/ghostery/user-agent-desktop
https://www.ghostery.com/

"Firefox and Chrome have been found to continuously write so much data that it might affect the durability of SSDs. Both browsers write an incredible amount of data to disk, even when idle.

firefox-with-32gb-written-in-a-single-day", this was in 2016

can't sign in to Firefox sync account (windows)

Says muted by you (not true) and does not do anything when I click to join the VC. This is default profile, nothing changed.

Hey, looks like you uploaded the wrong linux build for 115.4.0

LibreWolf is available as a Flatpak, as is standard Mozilla Firefox.

Scrolling down & looking at the number of downloads these browsers receive, LibreWolf gets consistently over 650 downloads per day & Firefox gets over 7,000 per day. Mercury is incredible, and I think the average user would enjoy its benefits. Purely from an adoption standpoint, I could see Flatpak distribution doing great things for Mercury. Many distros are configured with Flatpak support out of the box now, and it would make Mercury far more accessible to the average user. I really hope you consider this, as I'd love to use a Mercury Flatpak & I'm sure many others would as well. Thank you!

Please release 115.3.1 which fixes a critical security bug.

https://web.whatsapp.com/ doesn't work with this error on archlinux with the latest aur release.

Mercury gets 16.7 on https://browserbench.org/Speedometer2.1/
firefox gets 204
thorium avx gets 246
throium avx2 gets 218
edge gets 212

it's noticeable in regular browsing which is why I went to benchmark everything... and then saw it must be some bug?

I think we should integrate Betterfox because it has Speedfox and smoothfox that helps more smooth experience, also it is not strict as much as arkenfox's userjs so it is more suitable for daily users.

WhatsApp works with Mozilla Firefox 60+

Hello, for more smooth playback, we hope that these features are enabled in our beloved Mercury Browser please ? or may be it is good idea to add option to enable these flags manually....

Experimental MPEG-DASH support

and

HEVC/H.265 support on Windows

Thank you !

Can't live chat on youtube

Hello, I am experiencing bad Speedometer performance compared to standard Firefox on the latest Mercury 111 from the AUR.

➜ ~ inxi -v CPU: 16-core (8-mt/8-st) 13th Gen Intel Core i7-13700K (-MST AMCP-) speed/min/max: 1450/800/5400 MHz Kernel: 6.2.8-zen1-1-zen x86_64 Up: 17m Mem: 4548.1/31868.8 MiB (14.3%) Storage: 6.83 TiB (22.7% used) Procs: 499 Shell: Zsh inxi: 3.3.26

Troubleshooting Information (about:support)

Let me know if you need any more information. Thank you!

As mentioned in the Release notes, JPEG XL (JXL) support was added via mozconfig and --enable-jxl. However, it isn't enabled out-of-the-box as of release v.110.0a1:

Going to https://jpegxl.info/test-page shows:

As you can see, the JXL flag is not enabled by default:

Enabling the flag manually allows the JXL images to appear as normal. I think you need to change the user preference by default as is done by Waterfox (see 000-waterfox.js)

On a related note, I think you might also want to disable the Nightly check for JXL support as done in this Waterfox PR, in case in the future you switch to building on Stable. I'd open a PR myself, but I'm not sure how to integrate it within your repository.

Thank you!

Is there a reason why Mercury is based on Firefox ESR? Are you planning to support current Firefox versions?

Maybe it's because I had Pulse -Firefox fork- installed and it left something in one of the folders, even after uninstalling it?

On Fedora 38's default GNOME Wayland session, there is a modest performance gain on Speedometer 2.1

Xwayland

Wayland

It's not earth-shattering, but it's more or less "free" in the sense that Firefox has had great Wayland support for a while & I cannot think of any features missing/regressing other than PiP not automatically staying on top, and the GTK4 touchpad scrolling fix not being backported to GTK3 yet.

The version string seems to be 115.3.0esr, which leads me to believe that this is based off firefox 115.3.0, but the problem is fixed in firefox 117.0.1

When I right click files in a google drive folder, nothing happens and they don't download. I'm using the latest version of Mercury with extensions disabled. The same issue occurs whether I'm signed into Google or not. I didn't change any of the cookie settings, and have it set to 'Standard'.

I didn't notice any mention of these optimization techniques in the docs. Could we also apply them to squeeze out a few percentage points of performance? From what I've seen, BOLT can improve performance by 3% for binaries already optimized with PGO and LTO.

It would be awesome if the amazing WinUpdater could be expanded to include Mercury as well (or it's own fork if necessary)