-
WG9s
WARNING - do not use the currently posted firefox nightly it will mess up your profile
-
teruna
Hello! Just wanted to report that the latest SM can be compiled with clang 13.0.0 and rust 1.56.1 on Arch Linux.
-
WG9s
teruna: by latest I presume you mean 2.53.11b1pre? as far as I know 2.53.10 will still not build with rust 1.56.0 or later
-
teruna
WG9s: Yes, using the latest patch queue from your homepage.
-
WG9s
OK I was panning to talk to frg to see if we can create a zip file with the patches required to build 2.53.10 with 1.56.1. the patches were just finalized and tested yesterday and 2.53.10 was already built and ready for release afgter last minute things like registering the signed versions with firewall vendows so they will be whitlisted for making connetions to webservers
-
WG9s
I was thinking creating a zip file containing the patches and a series file or perhaps I could create a rollup file with all the patches combined into one.
-
WG9s
the firefox nightly issue is being trqcked in
Bug 1740987
-
WG9s
the fix has landed but nightlies were not retriggered
-
frg_Away
For what it is worth. If you want to compile 2.53.10 with rust 1.56.1 you need 1471096-1-63a1.patch and 1730397-4only-94a1.patch from
wg9s.com/comm-253/patches/seamonkey…253-patches/mozilla-release/patches .
-
frg_Away
1471096-1-63a1.patch will not apply clean and needs a mini rebase for toolkit\library\rust\shared\Cargo.toml
-
WG9s
)K I might be able to get rebased patchs and we can refernce this on the website for how to build from source
-
WG9s
ca either do a rollup patch including both or just rebased patches. ithis is something I can do
-
coldice
@wg9, if I want to build seamonkey 2.53.x on macos do I need any of those patches, no?
-
frg_Away
coldice has been some time since I natively built on macOS Probably just checkout mozilla-central and do a mach bootstrap to get the environment in shape. You donÄt need patches if you use 1.55. We build with the 10.11 sdk. Later one might need fixes.
-
frg_Away
Building native arm M1 will fail and can't be done yet.
-
tomman
frg_Away: hasn't Apple added a bunch of new restrictions on later macOS releases because they now think 3rd-party non-cellphone apps are malware?
-
tomman
from the notarizing stuff to being unable to take screenshots without invoking the complete security theater with obscure/barely documented APIs
-
frg_Away
tomman If we start to sign them we need to notarize too I think. So far it works but you need to manually override when you install from the internet.
-
frg_Away
With native arm applications notarization is mandatory I think. But not there yet.
-
tomman
And it's not only your app but also the entire build toolchain
-
frg_Away
tomman we are doing cross compiles like mozilla. Only the final binaries need to be notarized eventually.
-
tomman
that sounds saner
-
coldice
@frg_Away, I'm targeting ia-32/x64 :)
-
frg_Away
coldice Well x64 should work but not well tested. At least not by me :) If you run into problems or need a backport for a bug to continue let me know.
-
coldice
Great, many thanks!
-
frg_Away
coldice got the 10.11 sdk from here:
github.com/phracker/MacOSX-SDKs
-
frg_Away
I know that some later patches are missing where you need to put the sdk patch in the mozconfig. So optional but I am sure latest 10.15 or newer will not work out of the box.
-
frg_Away
coldice my core cross mozconfig but a bit older
paste.debian.net/1219415
-
frg_Away
We recently switched to a later clang but more or less the same. Not sure how this translates to a native compile.
-
coldice
Many thanks for you input, @frg_Away , appreciate it!!
-
mlh85386
Is it possible to build defined branches of the source tree and test a binary whether it needs the SSE2 instructions or not? From building SM 2.53.1 I know that Flags for gcc avoid usage of these instructions. But I do not have trust in the rust compiler. I already set RUSTFLAGS='-C target-cpu=k6'. Nevertheless target "i686-unknown-linux-gnu" appears in the mach's configuration log for...
-
mlh85386
...rustc/cargo.
-
mlh85386
Does somebody know whether there is a package of rust programm files available to do a fast test whether the rust compiler honours the option -Ctarget-cpu=k6 or not?
-
mlh85386
good evening, good bye