You can hide them with userContent.css - most of the devtools window stuff is styled via userContent.css not userChrome.css.

But there’s a catch.

Browser toolbox is essentially a separate instance of Firefox, running in separate profile so your “normal” user css files don’t apply to it. Thus, you need to first enable the toolbox profile to load it’s own user css files and create them just like you do normally (toggle toolkit.legacyUserProfileCustomizations.stylesheets, create files in chrome/ folder etc.). The toolbox profile is stored inside the regular profile - in a directory chrome_debugger_profile.

To get to about:config of the toolbox profile you need to first open a new main-window for it - one way that works is to click the meatball menu at the top-right of the toolbox window, and select “Documentation…” - that will launch a new window using using the toolbox profile and then you can just open about:config and proceed as usual. Or you can just modify prefs.js of the toolbox profile directly while the toolbox is not running.

Anyway, after you have set up the toolbox window to load user css files, then just slap this to its userContent.css and restart the toolbox:

header.chrome-debug-toolbar{
  display: none !important;
}

Yeah, !important doesn’t affect inheritance in any way. It only means that this particular rule is to be used if there are multiple rules that would set that particular property for the element in question (unless there’s some other more specific rule with !important tag as well). MDN lists property inheritance in the formal definition section. You can totally make background-color inherited though - like *{ background-color: inherit } (and then set the property to something else on the root element from which you would want to inherit from) but it would then either not apply if website set it to anything else for an element or override any other set value if you used !important tag.

One other thing worth noting is that I would not recommend the rules mentioned for userChrome.css to be used as is - at least on Windows they completely break Firefox startup - it fails to display any window if you do that. Instead you should add a [sessionrestored] selector to wait a bit before those rules are applied to main-window:

#main-window[sessionrestored], #tabbrowser-tabpanels{ background: transparent !important; }

Right, background-color is not an inherited property (compared to for example color (color of text) which is). But even if it were, inheritance is not “enforced” so if website css sets a backround-color specifically for that element then the inherited value would be lost anyway.

But the way you now describe it doesn’t seem possible. There is not syntax to apply style rule to “just the innermost element”. I think the closest would be to have everything else have fully transparent background, but the html root element have only partial transparency:

*{
  background: transparent !important;
}
html:root{
  background: #00000080 !important;
}

However, you will still face a problem; many websites draw graphics or images as a background-image so if you use the background shorthand property then those graphics will be effectively removed. On the other hand, if you instead set just background-color then parts might get opaque again because a website could be drawing even opaque backgrounds as background-image instead of background-color.

I think the answer depends on which elements exactly you want to make transparent. The page is a layered structure. The html root element is behind them all. Then body element is on top of that, the rest of the elements on top of body, etc.

So if you intend to have transparency all the way down, then you need to make sure that all the elements in that stack are transparent. If any single item in a stack has an opaque background then the transparency effect stops at that.

As an example, if you set background:transparent to just body but not the document root element, then body will indeed be transparent, but it does not matter because the root will still be opaque. Likewise, if root is made transparent, but there is any opaque layer on top of that, then only the parts of the root element that are not covered by that opaque layer will show up as transparent. If you have a glass table and put a sheet of paper on top of it, then you don’t see through the part covered by the paper even though the table itself is transparent.

Yeah, I just figured the safest option would be to only set the actual document root element transparent - in practice I think it’s possibly more likely that the <body> element has background set by the page - although the page might as well set both. So yes, it depends on the website.

I don’t think I understand exactly what parts you want to make transparent, but this does work:

1. set browser.tabs.allow_transparent_browser to true
2. in userChrome.css add #main-window, #tabbrowser-tabpanels{ background: transparent !important; }
3. in userContent.css add html:root{ background-color: transparent !important; }

The above would make window background, and the are behind web-content transparent as well as background of html documents - otherwise the background of browser area wouldn’t show up anyway. Toolbars that have their own specified colors would still be colored - which might be opaque or not depending what theme you have selected.

Would be pretty idiotic to not close it, otherwise opening a bookmark would always require a second click to close the popup.

Anyways, you can go to about:config and set browser.bookmarks.openInTabClosesMenu to false - afterwards you can hold Ctrl (or just click the middle mouse button) while clicking a bookmark from such popup and the popup should stay open.

But if you’re using the built-in auto-updater (like people tend to do on Windows and macOS), then it happens automatically in the background, unless you tell the auto-updater to not update automatically.

Definitely does not work that way on my Windows 10 installation. When update is available, Firefox will have a “Restart to install updates” in menu button notification - but the files are not replaced on disk until you actually close (or restart) Firefox and thus Firefox continues to work normally.

What can happen though is that if you run another instance (ie. another profile) of Firefox while the first one has “staged” the update then that another instance can trigger the files to actually be replaced on disk but you would very deliberately do that.

Firefox shouldn’t force you to restart and update like that unless something else, such as your package manager, has already replaced the executable files on your disk. In such a scenario Firefox doesn’t have any option except to inform you to restart it (well I guess it could choose to crash). But the mechanism that forced the update is the package manager.

Felix Mikolasch, data protection lawyer at noyb: “Mozilla has just bought into the narrative that the advertising industry has a right to track users by turning Firefox into an ad measurement tool. While Mozilla may have had good intentions, it is very unlikely that ‘privacy preserving attribution’ will replace cookies and other tracking tools. It is just a new, additional means of tracking users.”

Sigh… I cannot for the life of me figure how anyone could think that enabling PPA (even by default) means that advertising industry has somehow right to track folks. Like dude, the entire point of PPA is that advertisers could then get to know if/when their adverts are working without tracking people.

The argument that “It is just a new, additional means of tracking users” also doesn’t really make sense - even if we assume that this is new means of tracking. I mean, sure it technically is new addition, but it’s like infinity+1 is still infinity - it doesn’t make a difference. The magnitude of this one datapoint is about the same as addition of any new web api (I mean there are lots that shouldn’t exist - looking at you chromium… but that’s besides the point).

File a complaint over use of third-party cookies and actual tracking if you want to be useful - this complaint just makes you look like an idiot.

Also, Servo was originally more or less a testbed for new rendering pathway (webrender) which, when ready, was then integrated into Firefox.

True, and I agree - for this feature to be effective the site-specific rules need to be maintained properly.

All I’m saying is that it leaving some query parameters unremoved is not indicative of the feature not working. If you want to add more query parameters to the removed list then feel free to open a bug about it.

That feature removes parameters that are known to be used for tracking. It does not remove all query parameters willy-nilly. For example on youtube it should remove si, feature and kw parameters as well as a set of parameters on a list that applies to all websites. However, pp parameter is not in that to-be-removed list.

As an example v parameter is for video id on youtube, it would be kinda silly if that was removed, so the feature kinda has to do some site specific action.

Reader mode can show images as well though. I mean, it isn’t always successful in doing so - probably because there’s about a million different ways a random website could show an image - but reader mode can show images that it thinks are part of the actual content.

I’m not seeing any such issue with Nightly on my Fedora system.

Indeed. Also, the auto-open PiP is super nice feature I’ve been using a lot.

I don’t think Firefox Color can do that. At best you could test and set colors using Color, and then export the settings for both as a theme .zip or .xpi files, and then combine the two to a single manifest.json file. Inside the manifest, a "theme" key would be color properties for “normal” theme and "dark_theme" would be for dark-theme. After that you would submit the theme package to get it signed after which it can be installed as a real theme.