How is PaleMoon 5 times smaller than Firefox on macOS? Topic is solved

[marigold]

I use PaleMoon as my regular browser and also have Tor Browser (Firefox) as my stand-by browser. While upgrading PaleMoon to 34.3.0.1, I observed that the latest version of Palemoon is only around 103+ MB where as the latest version of Tor Browser (Firefox) is around 503+ MB - that's 5 times bigger than PaleMoon! Even though I had deleted the Tor module (around 70+ MB) from the Tor Browser, I thought maybe the large size is just particular to the Tor Browser. So I downloaded the latest version of Firefox (macOS), and to my surprise, that is also around 503+ MB.

Now, I know that while PaleMoon is a fork of Firefox its codebase has diverged from it. But even without the code for something like WebRTC, WebExtensions and DRM media extensions, is there any particular technical decisions that have made PaleMoon / Goanna 5 times smaller than Firefox / Gecko?

(Kudos to the PaleMoon team for that)!

[BenFenner]

I'm way too lazy to check, but it is unclear if you've confirmed the size of the Pale Moon browser installed on disc, or the installer size itself, or only the size of the update? Those are all potentially 3 very different sizes (not to mention the size of the codebase) and each only so meaningful on their own. Then you're trying to compare that to what exactly? Other browser installer sizes?

Don't make us work for this.

[jobbautista9]

I think Firefox's macOS version is a "fat binary", which allows it to support both Intel and ARM in the same package. We have separate DMGs for each architecture.

[marigold]

... it is unclear if you've confirmed the size of the Pale Moon browser installed on disc, or the installer size itself, or only the size of the update? ...

You are right - I should have highlighted that I was talking about the installed size of both applications.

[marigold]

I think Firefox's macOS version is a "fat binary", which allows it to support both Intel and ARM in the same package. We have separate DMGs for each architecture.

Yes, that seems to be it.

Code: Select all

> file /Applications/Pale\ Moon.app/Contents/MacOS/palemoon

Code: Select all

/Applications/Pale Moon.app/Contents/MacOS/palemoon: Mach-O 64-bit executable x86_64

Code: Select all

> file /Applications/Tor\ Browser.app/Contents/MacOS/firefox 

Code: Select all

/Applications/Tor Browser.app/Contents/MacOS/firefox: Mach-O universal binary with 2 architectures: [x86_64:Mach-O 64-bit executable x86_64
- Mach-O 64-bit executable x86_64] [arm64:Mach-O 64-bit executable arm64
- Mach-O 64-bit executable arm64]
/Applications/Tor Browser.app/Contents/MacOS/firefox (for architecture x86_64):Mach-O 64-bit executable x86_64
/Applications/Tor Browser.app/Contents/MacOS/firefox (for architecture arm64):	Mach-O 64-bit executable arm64

Assuming that the individual binaries of Firefox, for the two platforms, are roughly half the size (i.e. around 250 GB) that is still commendable for PaleMoon ...

[Moonchild]

It's because of their massive bloat due to Rust, if nothing else. Their xul.dll in Windows is currently a whopping 172 MB in size (as opposed to our 38). I'm pretty sure most of that is due to Rust. Add to that a ton of gadgeteering support and "remote" control code and telemetry, and it adds up.
Not to mention a bunch of "helper apps" like a messaging proxy that tightly integrates with Chrome if installed on the same system.

[UCyborg]

Chromium's chrome.dll is 2x times that and that's still C++ I think. 🙃

[Moonchild]

Actually, looking at the nmhproxy included in Firefox, I fell into a bit of a rabbit hole. Neither the bz bug introducing it nor follow ups have descriptions, at all. And they link to Jira as a see-also that is restricted to Mozilla devs only. So basically, Firefox comes with a black box listening proxy that listens for native messages from Chrome that can directly control the browser... !
I'd be extremely weary of such a setup. That would be an easy way for malware to hijack Firefox if they figure out the control protocol for it (seems to be json from stdin)...

[moonbat]

Chromium's chrome.dll is 2x times that and that's still C++ I think. 🙃

Chrome wants to become a VM for web applications and are probably trying to replicate functionality that belongs in an operating system.

[UCyborg]

^ You could say it already is.

[Moonchild]

Chrome wants to become a VM for web applications and are probably trying to replicate functionality that belongs in an operating system.

^ You could say it already is.

Pretty much. And Chrome also controls the specs of that VM.
We're partially there as well with how our JS has massively expanded. Unfortunately it has its drawbacks as well.

[marigold]

Not to mention a bunch of "helper apps" like a messaging proxy that tightly integrates with Chrome if installed on the same system.

Why would Firefox need "helper apps" to integrate with Google Chrome? Perhaps it is for "syncing browsing data" between the two - if I remember right, both can can import bookmarks and maybe even history from each other? But it is weird if, as you say, Google Chrome can send messages to Firefox and ... do something ... !!? (Or did you mean the "Firefox's chrome"?)

[Moonchild]

No i meant the Chrome browser. And no, it is not for "syncing data" it is literally an "in" for a native messaging protocol to launch and control parts of Firefox. That has zero actual documentation open to the public.

[Gemmaugr]

Does this shed some more light on that black box? https://bugzilla.mozilla.org/show_bug.cgi?id=1876342

[Moonchild]

Does this shed some more light on that black box? https://bugzilla.mozilla.org/show_bug.cgi?id=1876342

Not really, as it just deals with system registration of the proxy as a scheme handler; it doesn't explain anything about the proxy itself other than a drive-by mention of a "dual browser" setup and launching Firefox from within Chrome, which I already figured out before. It says nothing about why this would be a thing to begin with, or why Firefox needs to integrate so tightly with Chrome, or how it can be used in good or bad ways.
Assuming Firefox defaults to accepting known extension IDs in Chrome, I can see abuse being easy.