e10s / Electrolysis

[LimboSlam]

UnloadTab as it offers a working whitelist option and is the most stable of all those dead suspend/unload tab add-ons. This one should really be converted to a Pale Moon specific one.

Maybe I'll be the one to do it once I'm fully confident of my coding skills (this thread was a great influence to keep at it with coding: viewtopic.php?f=5&t=11529&start=40#p81347). But hopefully someone else with the experience can do it sooner and better.

[snertev]

Stability of e10s 45 beta is a lot worse than no-e10s 45 beta, according to FF telemetry.

Data taken from here:

http://nbviewer.jupyter.org/github/viti ... ysis.ipynb

type of crashes - a crash every N minutes of global usage (no-e10s)/a crash every N minutes of globl usage (e10s)

chrome crashes - 1 every 4,44m/1 every 8.47m -
content crashes - 1 every 27,52m/1 every 3.75m -
plugin crashes - 1 every 7,70/1 every 5.09m -

TOTAL (any type of crash)

no-e10s: 1 crash every 2.55 minutes of global usage
e10s: 1 crash every 1.73 minutes of global usage

Here:

http://nbviewer.jupyter.org/github/viti ... outaddons/

there are other data for responsiveness and plugin lag, both slightly better in no-e10s (responsiveness better in no-e10s... OMG!!!)

A lot of work for nothing, so far.

[Moonchild]

A lot of work for nothing, so far.

"I told you so"

[snertev]

A lot of work for nothing, so far.

"I told you so"

I know, I know...

The decision to develop a e10s Firefox version is simply wrong and proved by facts: less stability, less responsiveness after more than 2 years of development.

They should have developed the Servo thing while maintaining the Gecko browser as it is. Once ready, the venerable Gecko core could have been set into a legacy state for some time and then retired, at least as far as Mozilla foundation is concerned.

Indeed, they have literally wasted years of development for nothing.

So, the facts show me that till it will be a viable browser in current web, Pale Moon is still the wiser choice.

When the dust of Mozilla's disaster will have settled, within a couple of years, the re-based Pale Moon may still be the saner survived open source browser, who knows...

[Moonraker]

personally i feel mozilla are attempting to implement a function into a browser which was not initially designed for such a function.Only way mozilla could make that work would possibly create a new browser from scratch.

They are trying to put a square inside a circle as we would say.

[snertev]

personally i feel mozilla are attempting to implement a function into a browser which was not initially designed for such a function.Only way mozilla could make that work would possibly create a new browser from scratch.

They are trying to put a square inside a circle as we would say.

You're absolutely right. Microsoft and Opera tried to start from scratch without touching the old browser. Like it or not, that was the way to go.

Unfortunately, Mozilla is in the middle of a river: the new browser is not even functional yet and their user base is abandoning ther ship. So: "let's change that old code and annoy at death the faithful user base".

This happened because they thought with their corporate "brain" (Panic! Our user base share is dropping! We'll have less money!) rather than with their foundation "brain" ("We have to innovate the web. Period.").

It's the same mistake made by Opera, though they were not a foundation, but just a savvy norwegian tech corp.

[BaronHK]

So uh... Moonchild, could you answer the first question?

Compiler parallelization, pretty much. Or rather: properly changing code so it can be executed in parallel, then letting the compiler take advantage of it.
I've previously looked into using OpenMP but the implementation of that isn't particularly good in MSVC and seems unmaintained, so I've moved to the new built-in parallelizer instead, which has been in use for quite some time. See also: https://msdn.microsoft.com/en-us/librar ... 10%29.aspx

Hi there. I'm sure you've thought about this, but I think that GCC probably has better OpenMP support than MSVC, so have you considered cross compiling the Windows build of Pale Moon to take advantage of GCC?

[BaronHK]

personally i feel mozilla are attempting to implement a function into a browser which was not initially designed for such a function.Only way mozilla could make that work would possibly create a new browser from scratch.

They are trying to put a square inside a circle as we would say.

You're absolutely right. Microsoft and Opera tried to start from scratch without touching the old browser. Like it or not, that was the way to go.

Unfortunately, Mozilla is in the middle of a river: the new browser is not even functional yet and their user base is abandoning ther ship. So: "let's change that old code and annoy at death the faithful user base".

This happened because they thought with their corporate "brain" (Panic! Our user base share is dropping! We'll have less money!) rather than with their foundation "brain" ("We have to innovate the web. Period.").

It's the same mistake made by Opera, though they were not a foundation, but just a savvy norwegian tech corp.

Opera got significantly less usable when they switched to Chromium. They tossed the M2 mail client, their Opera Unite feature, the IRC support, the note-taking support, and so on, and they became, oh, more or less another Chrome clone. Different branding.

It seems that everyone is throwing in with Webkit/Blink because it's easier that way and ignoring that it doesn't have all of the unique features that Firefox has. But it seems that Mozilla is (1) not valuing the unique features that Firefox has, (2) wasting time with e10s, and (3) planning to throw those unique features away and implement Chrome-style extensions instead. Yuck.

[Moonchild]

but I think that GCC probably has better OpenMP support than MSVC, so have you considered cross compiling the Windows build of Pale Moon to take advantage of GCC?

GCC is also vastly inferior for compiling Windows applications, especially those of this magnitude, than the Microsoft compiler is (perfectly logical, too, considering the compiler and target are both from the same vendor). Better OpenMP support than what MSVC offers is going to be completely overshadowed by the basic compilation quality when GCC would be used. I don't see a reason to toss out the best tool for the job in favor of one that happens to offer better support for one of the possible ways to parallelize at the compiler level.

[New Tobin Paradigm]

Here is my perspective on this:

Cross-compiling to Windows from Linux (or using gcc) is moot. While support for building Windows applications with GCC has purportedly gotten better in recent years there is no real way to prove that it would be up to dealing with such a complex codebase such as Pale Moon since cross-compiling to Windows is wholly busted in the build system and hasn't been properly supported for MANY years maybe not since the old Netscape days.

While it remains theoretically possible there is a bit of a catch-22 situation where this is concerned. We don't know if it is worth expending the effort to fix cross-compiling in the build system without knowing if such a build would be worthwhile and we can't get a build to test to see if it is worthwhile without fixing cross-compiling.

Basically this all just boils down to if we should support more than one defined and least painful way to compile for a specific platform. I am not sure it is worth all the extra trouble. For now, Linux target is built on linux with gcc, Windows target is built on Windows with msvc, and Mac target is built on mac with whatever it uses.

tl;dr RESOLVED WONTFIX