On 2021-02-25 4:40 p.m., Randy MacLeod wrote:
Richard,
I won't have a v3 today.

I'm adding some of the people have worked on meta-rust as an FYI.
Sorry that I haven't been CCing y'all earlier.

If you'd like to be dropped from the thread say so on list or to me privately.

One brief comment below but I haven't had much time
to work on this today yet.


FYI, the Rust merge to oe-core was discussed at yesterday's monthly
Yocto technical call and there was unanimous agreement that
we would NOT merge Rust in 3.3 but rather wait until the next release.

We're waiting because there are still a few bugs to be resolved
and so far only librsvg is blocked on the lack of Rust. When we
uprev python-cryptography, that will also need Rust so we really
need to get the known problems fixed in the coming weeks so
that we can be ready to merge early in 3.4. That will leave us
plenty of time discover any new issues and then have a well-tested
release including Rust in the autumn of 2021.

I'll continue merging patches from meta-rust and publishing updates
in my poky-contrib branches every week or so until 3.4 opens up.


Thanks everyone,
../Randy



On 2021-02-25 12:37 p.m., Martin Jansa wrote:
On Thu, Feb 25, 2021 at 04:55:43PM +0000, Richard Purdie wrote:
On Wed, 2021-02-24 at 20:48 -0500, Randy MacLeod wrote:
Added or updated these commits:

    86109491bc maintainers: Add myself as maintainer for rust pkgs
    ...
    b411e322b3 rust: remove container build scripts used by meta-rust
    c646d47ebc rust: update the README to conform to being in oe-core
    e19d823567 meta-rust: merge commits

The last commit update was just squashing a new commit from meta-rust:
    f8b9103 Remove checks for Rust versions we don't build anymore.


To Do List
==========

1) cargo-bitbake - the README suggest adding this using desktop
    cargo but once we get the SDK merged, it should be added there.
    Someone suggested using devtool for this rather than cargo-bitbake.

2) Fix the SDK libcrypto, libstdc++ glibc-2.33 problem. I'm running a
git bisect using poky, meta-oe/master, meta-rust+SDK patches.

3) Add a test case to either:
    - just build rust-hello-world or
    - run it in qemu

N) Fix *some* of the remaining bugs in:
       https://github.com/meta-rust/meta-rust/issues
I put this into master-next to see how it looked. The build hasn't completed
yet but so far:

rust-native doesn't build on debian8:
https://autobuilder.yoctoproject.org/typhoon/#/builders/116/builds/40/steps/13/logs/stdio

oe-selftest:
https://autobuilder.yoctoproject.org/typhoon/#/builders/86/builds/1855/steps/14/logs/stdio
https://autobuilder.yoctoproject.org/typhoon/#/builders/87/builds/1887
(others will likely show the same issues)

which is basically three tests failing:

oe-selftest -r sstatetests.SStateTests.test_sstate_noop_samesigs
oe-selftest -r sstatetests.SStateTests.test_sstate_sametune_samesigs
Not sure how to read these, but it might be fixed in upstream by:
https://github.com/meta-rust/meta-rust/commit/920328c8d8dd553e69309733d874229af4fa159f the sstate-diff-machines.sh doesn't show with an issue for rust-native with this one applied.

And it did show an error without it:
  === Comparing signatures for task do_populate_sysroot.sigdata between qemux86 and qemuarm === ERROR: rust-native different signature for task do_populate_sysroot.sigdata between qemux86 and qemuarm
NOTE: Starting bitbake server...
NOTE: Started PRServer with DBfile: OE/cache/prserv.sqlite3, IP: 127.0.0.1, PORT: 46469, PID: 57036 Hash for dependent task rust/rust_1.49.0.bb:do_install:virtual:native changed from 7334d76a4757c1b893a8c2b8c85ada005788b3adeff5a33fe509c43f994d91c1 to 162ccf6462851630b0816fdb5fbe68f6b158b81f4ce3ed45ec4c1f258f93212c Unable to find matching sigdata for virtual:native:OE/meta-rust/recipes-devtools/rust/rust_1.49.0.bb:do_install with hashes 7334d76a4757c1b893a8c2b8c85ada005788b3adeff5a33fe509c43f994d91c1 or 162ccf6462851630b0816fdb5fbe68f6b158b81f4ce3ed45ec4c1f258f93212c

ERROR: 1 errors found in BUILD/sstate-diff/1614274412/signatures.qemuarm.do_populate_sysroot.sigdata.log

I have that in my patchset locally:

commit e19d82356755a42a46221cb1a0fbacf6339560a1
Author: Randy MacLeod <[email protected]>
Date:   Tue Feb 23 21:26:23 2021

     meta-rust: merge commits

     Merge most of:
       git log --oneline --no-merges 5cda04c..029f1ea
     with two layer.conf exceptions explained below.

     f8b9103 Remove checks for Rust versions we don't build anymore.
     920328c rust-native shouldn't depend on TARGET variables

and I confirmed that it's in master-next.


../Randy


oe-selftest -r distrodata.Distrodata.test_maintainers

so the maintainers entries aren't quite right which is easy to fix and
the sstate signatures aren't behaving correctly which may be more fun,
depends what is wrong. Those tests explain they're broken (which is what
we have the tests for) but more investigation would be needed to
understand how/why.

Cheers,

Richard










--
# Randy MacLeod
# Wind River Linux
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#148921): 
https://lists.openembedded.org/g/openembedded-core/message/148921
Mute This Topic: https://lists.openembedded.org/mt/80893251/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to