Re: Debian sid testing progress
On 08/14/2018 09:54 PM, Riccardo Mottola wrote:
> there is a lot of good stuff, but also hold your breath. I am working on the code to generalize his work so that it works on x86-32bit (64bit is currently
> off-limits for the code, it would need even more work to remove Carbon) so that it can be used on older intel macs too. It is absolutely not an easy task, I
> have been working closely with Cameron for months: his code was worked for years against a single OS and CPU target. So in my case it had endianness issues,
> which are to a larger extent fixed.
Did you port the WebAssembly stuff to PowerPC? Because that's actually one of the
things that keep Firefox from building on 32-Bit PowerPC. For 64 bit, it doesn't
seem to need any additional code, I just had to add a few lines upstream to get
sparc64 to build, for example.
> We have a x86 build working as a proof, but I did not release anything to the public since it is not as stable as the PPC version, although it proves how much
> can be "squeezed" from older Macs.
On which Firefox version is this based?
> By analyzing the code, I can tell you that a lot would be needed to get them on non-Mac systems. Of course, it could be done.
Not sure what should be Mac-specific here? The JIT and WebAssembly stuff is not
platform-specific, just arch-specific - as far as I know.
> I'm not sure Quantum is better in anything. On NetBSD I am using the rust-based Firefox and it appears more bloated and trade-off memory with performance. Also,
> I personally dislike the "new" interface, but that is taste.
> Rust is a pain and slow to death, slower actually. I build it on x86 machines which are faster than all G4 machines I know and it takes hours/days. We already
> have pain with gcc and clang... now rust, oh, man.
Rust is unfortunately currently broken on 32-Bit PPC and I have been bisecting
the bug for days now: https://github.com/rust-lang/rust/issues/50960
Bisecting Rust is a completely pain in the as since the git history is completely
non-linear.
> Perhaps Pale Moon could be ported to PPC? The community is not that happy with ports/forks apparently though, although it appears to be only a naming issue and
> an issue of dependent libraries.
PaleMoon should just build fine on PPC since it's just an older Firefox. In fact,
I actually submitted some patches to PaleMoon upstream to fix the build on m68k
(although I think it's missing some alignment patches) ;).
I have more patches for PaleMoon in the queue to fix the build on sh4, for example.
Adrian
--
.''`. John Paul Adrian Glaubitz
: :' : Debian Developer - glaubitz@debian.org
`. `' Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
`- GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Reply to: