Development direction feedback

[andyprough]

That's a very small number. Maybe 0.5% that actually use the current version of Pale Moon on Windows 7? I would have thought it was much higher based on the number of Windows 7 users who post on the forum, more like 10% or more.

It just shows you that most Pale Moon users aren't very vocal about it or don't feel the need to be active on the forum.

If those stats are correct then it's possible that most Pale Moon users don't report issues on the forum because they are using the current version on a targeted current OS where Pale Moon works very well. That's my experience anyway. I often notice that I don't experience issues that are raised on the forum by others, especially with respect to slowness, stuttering of video, scrolling problems, etc. Might also have to do with the fact I mostly use hardware that's about 4 years old or newer.

[gepus]

I often notice that I don't experience issues that are raised on the forum by others, especially with respect to slowness, stuttering of video, scrolling problems, etc.

Neither do I. Hardware might be a factor for sure.

I'm one of the 1% using always the latest Pale Moon (manual update) on Win7-x64.
BTW, Pale Moon (-x64-avx) works fine as well since my CPU has support for avx (unfortunately too old for avx2 support).

[LAR Grizzly]

I'm one of the 1% using always the latest Pale Moon (manual update) on Win7-x64. My CPU has support for avx (unfortunately too old for avx2 support).

Ditto! (although, I use the internal updater)

[Gianluca18092004]

It is because even Windows 10 will become "unsupported" (by MS) next year.

We didn't discontinue Windows 7 support when that went EOL, though. The point isn't that we need to discontinue support for an operating system as soon as the EOL hits, but rather that it's reaching a point where people expect us to support operating systems for decades past their EOL. They want more than 5 or 6 years of legacy support, they want 10 or 20 years of it. To the point that critical libraries or compiler updates would require a newer OS. That's where we're trying to draw the line here.

Ok, but why is OS X Lion supported (until I read in 33.0 that experimental PowerPC will be added), but not Vista or XP? What is the difference between Vista and 7, or the big obstacle?

[R3n_001]

I still haven't found any difference going from SSE2 through to AVX2. I was originally going to have an AVX or AVX2 build of my own Firefox fork on top of SSE2, but in testing I found no quantifiable difference and just went for SSE2 x86 and SSE2 x64. Even better, UXP performs notably better compiled as x86 instead of x64 in my testing. Moving UXP to AVX makes no sense unless there is some measurable performance gain that I have not been able to find in UXP and even the latest browsers.

https://board.eclipse.cx/viewtopic.php?t=669

Ok, but why is OS X Lion supported (until I read in 33.0 that experimental PowerPC will be added), but not Vista or XP? What is the difference between Vista and 7, or the big obstacle?

There is none. UXP is a 52 ESR base, and 52 ESR natively supported XP. The XP UXP forks perform better in my testing even.

[moonbat]

There is none.

If I'm right, the issue is lack of compiler support for those OSes. Once Visual Studio stopped support for XP/Vista, Pale Moon had to do the same, and this was after the OSes themselves were EOLed.

[R3n_001]

You don't exactly need the latest of compilers. From what I've seen looking around spaces, older ones were usually faster and lighter. Not sure how much it applies here, but I was talking with a developer of another project and they said GCC4.7.2 compiles their project 5x faster than GCC8.4.0.

[Moonchild]

There is none.

If there was none, then I wouldn't have drawn the line between Vista and 7, and would rather have drawn the line between XP and Vista. Vista is, however, in a weird hybrid state between NT5 and NT6, and is missing some essential bits in the kernel for proper NT6 class support.

UXP is a 52 ESR base, and 52 ESR natively supported XP.

UXP was ESR52 base back in 2014, but even the first release of 25.0 already stopped supporting XP.
Got my dates and versions mixed up as it's been so long. First UXP version building on an ESR52 fork was actually 28 from 2018. And no, actual XP support in the code was already removed from v27 2 years prior.

The XP UXP forks perform better in my testing even.

They probably perform better because they have a cut-down feature set. We've been losing performance just like every other browser on the market as the already obscene web standards are just bloated with more bells and whistles to make that "the browser is a platform" pipe dream from a decade or so ago come true. You may want to ask yourself why you're using Pale Moon if other browsers perform better for you - I'm still a proponent of using the tools that best fit your wishes and workflow. If that's a UXP fork, then by all means, use it!

[moonbat]

GCC4.7.2 compiles their project 5x faster than GCC8.4.0.

GCC is irrelevant when we're talking about Windows,for which Pale Moon has always used some form of Visual Studio. There was a move to a newer version of VS sometime back, that offers better optmizations and it doesn't support older Windows versions.

[Moonchild]

There was a move to a newer version of VS sometime back, that offers better optmizations

Not just that, it also allows us to use necessary improvements in the C++ language - one of the few standards that are actually standards.

[dbsoft]

Ok, but why is OS X Lion supported (until I read in 33.0 that experimental PowerPC will be added), but not Vista or XP? What is the difference between Vista and 7, or the big obstacle?

This is or may be possible because modern GCC is now supported on PowerPC. The problem with XP is the modern toolchain is not supported there.