Package Details: veloren 0.16.0-2

Git Clone URL: https://aur.archlinux.org/veloren.git (read-only, click to copy)
Package Base: veloren
Description: The last stable release of an open-world, open-source multiplayer voxel RPG.
Upstream URL: https://veloren.net/
Keywords: game rpg veloren
Licenses: GPL3
Submitter: LunarEclipse
Maintainer: LunarEclipse
Last Packager: LunarEclipse
Votes: 23
Popularity: 0.006016
First Submitted: 2019-06-23 23:12 (UTC)
Last Updated: 2024-08-27 13:30 (UTC)

Dependencies (17)

Required by (0)

Sources (1)

Latest Comments

1 2 3 4 5 Next › Last »

tuxayo commented on 2024-07-16 20:10 (UTC)

Oh no, even with the earliest toolchain having the different linker (2024-05-18) there are still code incompatibilities :(

   --> /home/victor/.cargo/git/checkouts/shred-a3307cb74879c8e8/5d52c6f/src/meta.rs:402:38
    |
402 |         let invalid_ptr = core::ptr::invalid_mut::<R>((self as *mut Self).addr());
    |                                      ^^^^^^^^^^^ not found in `core::ptr`



For more information about this error, try `rustc --explain E0425`.
error: could not compile `shred` (lib) due to 1 previous error

tuxayo commented on 2024-07-15 19:02 (UTC) (edited on 2024-07-15 20:48 (UTC) by tuxayo)

setting options=('strip' '!debug') - it seems the debug build was taking all the memory.

Nice find. When reviewing makepkg.conf changes, I forgot it would increase memory and/or build time. The memory savings will be very very welcome.

and also export CFLAGS+=' -ffat-lto-objects' in build() to fix compiling ring

Good find, there is indeed an issue with ring, lto and gcc: https://aur.archlinux.org/packages/rsspls#comment-980121

https://aur.archlinux.org/packages/aichat#comment-959940

https://github.com/briansmith/ring/issues/1444


And there is hope! :) https://gitlab.archlinux.org/pacman/pacman/-/merge_requests/131#note_185966

https://blog.rust-lang.org/2024/05/17/enabling-rust-lld-on-linux.html

bonus: «Most binaries should see some improvements here, but it's especially significant with e.g. bigger binaries» (Heavy breathing)

On my next retry I'll change src/veloren/rust-toolchain to match latest dev version set by Veloren. https://gitlab.com/veloren/veloren/-/commits/master/rust-toolchain?ref_type=heads


@gdamjan:

and now it fails with: [...]

Oh no, hopefully the linker change from the recent toolchains will also fix this.

gdamjan commented on 2024-07-15 12:01 (UTC)

and now it fails with:

          /usr/bin/ld: /home/damjan/.cache/paru/clone/veloren/src/veloren/target/release/deps/veloren_voxygen-37616f1347406efc.veloren_voxygen.d42aca386f67380d-cgu.15.rcgu.o: in function `core::ptr::drop_in_place<shaderc::CompileOptions>':
          /rustc/098d4fd74c078b12bfc2e9438a2a04bc18b393bc/library/core/src/ptr/mod.rs:507:(.text._ZN4core3ptr44drop_in_place$LT$shaderc..CompileOptions$GT$17hb47594a3f6533e3eE+0x10): undefined reference to `shaderc_compile_options_release'
          collect2: error: ld returned 1 exit status

;/

gdamjan commented on 2024-07-15 02:41 (UTC)

What are your RUSTFLAGS in /etc/makepkg.conf?

all are defaults from pacman (RUSTFLAGS="-Cforce-frame-pointers=yes")

ok, I've fixed the issue by:

setting options=('strip' '!debug') - it seems the debug build was taking all the memory.

and also export CFLAGS+=' -ffat-lto-objects' in build() to fix compiling ring

tuxayo commented on 2024-07-14 22:41 (UTC) (edited on 2024-07-14 22:41 (UTC) by tuxayo)

@gdamjan was that the error you got?

In my case, no OOM in syslog and the build reports a build process getting a SIGKILL when that happen. I got a lot of OOM to learn that before putting enough swap ^^" I was watching memory usage and there where many GiB still free.

What are your RUSTFLAGS in /etc/makepkg.conf?

gdamjan commented on 2024-07-14 21:58 (UTC)

@tuxayo, well the OOM killer triggers, starts to kill all different processes

for ex. kernel: Out of memory: Killed process 242472 (rustc) total-vm:16324480kB, anon-rss:13832764kB, file-rss:2164kB, shmem-rss:0kB, UID:1000 pgtables:28772kB oom_score_adj:0

but it's not only that, other processes are killed too (because of low system memory)

tuxayo commented on 2024-07-14 03:11 (UTC)

@gdamjan What's your failure, bellow is mine: (I wish I had as much RAM, my breakout is like the mirror of yours: 11.5G RAM + 40G SWAP 💀💀💀 adding to the agony of my 2012 i5-3320M 🫠)

https://paste.chapril.org/?651d1c7a97c2da30#A1fWidLvk6VeA71Yxi56v6TEo5BFGUzeYQy9td8SKnea

I don't get what's really the error.

"/usr/bin/ld: final link failed: bad value" ?

RUSTFLAGS="-Cforce-frame-pointers=yes -C opt-level=3 -C target-cpu=native"

gdamjan commented on 2024-07-08 14:18 (UTC)

Did anyone succeed to compile the 0.16 version?

I have 32GB RAM and 16gb swap, but the compile destroys everything :)

LunarEclipse commented on 2023-08-12 17:09 (UTC)

@seqfault: I was a bit busy the past 2 weeks, but I've now forwarded your replies to people upstream so hopefully someone can help.

It might be worth opening an issue on gitlab (or asking on the discord server) so you can get more direct contact with them, as I'm quite clueless on what could go wrong here.

seqfault commented on 2023-08-12 16:44 (UTC)

@Mckol: still waiting