https://pact.io logo
Join Slack
Powered by
Hey <@UB7B75EE4>, when you have free time, please ...
# pact-rust
t
Hey @rholshausen, when you have free time, please help me review these PRs: • https://github.com/pact-foundation/pact-reference/pull/479 • https://github.com/pact-foundation/pact-reference/pull/462 Once merged, a new version for 
pact_ffi
(and probably another one for 
pact_matching
) would be great. Thanks
r
Yeah, no prob
t
There is an issue while compiling the binary: https://github.com/pact-foundation/pact-reference/actions/runs/12406909362/job/34636085258
could not compile 
pact_models
(lib) due to 255 previous errors; 2 warnings emitted
The log is quite long, I'm not sure it's my fault or not
r
Won't be your fault
Failed to generate pact.h, no idea why
Runs fine on my machine. I'll build the files and upload them
Oh, bummer. The release build has been changed, and it doesn't upload any of the artefacts if any step fails
I'll have to fix it tomorrow
y
doh!
good work Ron and Tien, nearly there :)
r
Ok, I have reverted the release build back to the previous working version, and it seems to have run OK
@Yousaf Nabi (pactflow.io) it is reverted back to your previous version, could you check that all the artifacts are as you expect? It looks OK to me.
y
shall do dudesterino
aarch64 macos binaries should drop the 
-apple-darwin
suffix
Copy code
libpact_ffi-macos-aarch64-apple-darwin.dylib.gz.sha256
sha naming has reverted was
Copy code
d4dc3cac6571d1bb1d0b8efb1d1260491a60a44e4999ff9f1ac6b8254cbb40c4 *libpact_ffi-linux-aarch64.so.gz
now
Copy code
fa6f556430bea8f88aaae939ad045edbd874a01da3d129745abcbbcced4c9dbf *../release_artifacts/libpact_ffi-linux-aarch64.so.gz
binary compat version looks good for aarch64 macos bins
Copy code
otool -l /Users/yousaf.nabi/Downloads/libpact_ffi-macos-aarch64-apple-darwin.dylib| grep -E -A4 '(LC_VERSION_MIN_MACOSX|LC_BUILD_VERSION)' | grep -B1 sdk
    minos 11.0
      sdk 13.0
that is because we explicitly set it in the script for m1 in the older script for x86_64
Copy code
otool -l /Users/yousaf.nabi/Downloads/libpact_ffi-macos-x86_64.dylib| grep -E -A4 '(LC_VERSION_MIN_MACOSX|LC_BUILD_VERSION)' | grep -B1 sdk
    minos 12.0
      sdk 13.1
let me check what change i think we are missing
wonder what went skewiff in the first instance! I always run the release flows on my fork to be sure when I do a CI refactor
we could just update shasums filenames and reupload, and rename the macos aarch64 binaries manually for this release and fix forward
r
Yeah, I reverted the commits Josh made, and thought it would go back to the state you had it at.
BTW, because GitHub dropped MacOS 12, the minimum sdk we can now use is 14.1
y
ahh okay
r
I don't know if that will impact things
y
I don’t think this change meant to be reverted https://github.com/pact-foundation/pact-reference/commit/7c09f177d2bf1e36bb9e53c55def6e09dc65b6a3 it does the renames
we can set a min deployment target env var (MACOSX_DEPLOYMENT_TARGET=11.0) and a later xcode will build backwards compat bins no prob (earliest for aarch64 is 11 - big sur)
r
Oh, the revert went back too far.
y
ps. I stil haven’t looked into why the build failed with that PR applied in the first instance
r
Ok, I'll revert the reverts, then revert that one.
y
man I remember doing my first revert
and not long after learning about revertrevert
makes me very happy that its called git, the git
r
Yeah, don't know either, it just blew up trying to generate the headers. But the crap thing about it is then generates no artifacts at all.
y
We could probably stop it from bailing out fully, as by that point we’ve already built all the binaries and might as well publish them! building headers is trivial to do after manually if required
split out step and have a continue-on-error so we still get ui feedback about the failure
r
Yeah, that was what I was going to do, and then saw there were no artefacts sadparrot
y
then you look at teh covfefe cup and go today is going to be a looong day
watching traveling-ruby build in gh actions under qemu because people want security of a build that has come from ci. me on the other hand with my ferrari of a laptop that I am using as effectively a thin client, is crying inside. just let me build locally and publish. we could stick the bit coin miners in CI builds and no-one would notice anyway XD
yo do I heard you like cpus. How would you like 0.x of a CPU. cloud maffs
r
I'm running a beast of a machine, 16 CPUs, 24 cores, 5GHz. And 23 of the cores sit idle running the browser while waiting for the CI build to complete.
Ok, I have reverted the repo back to the state it was in yesterday. I can't just revert that PR, it is causing conflicts, so I'll have to do each commit at a time.
Oh, stupid me. You have to revert backwards.
Ok, reverted all the commits and re-generated all the sha256 files. I'm calling this release dead to me.
t
Sound complicated. I will try to made some changes in PHP side to use this release. I will revert them if new release come out.
There is something wrong with the musl build. PHP simply can't load it, without any specific error:
FFI\Exception: Failed loading '...pact.so'
correct: only musl + arm64 has this issue
r
I'll try create a new release, but I'm gong to be on leave until the second week in Jan, so I would be able to do much more on this.
t
Everything work fine now. Thanks. Happy holiday!
Open in Slack
PreviousNext
Powered by