Conversation
[ci skip]
[ci skip]
[ci skip]
[ci skip]
[ci skip]
These are pending updates in agent migration stage 2.
CLI extracted from Polykey and migrated to Polykey-CLI
|
Looks like the mac and windows builds are failing in CI. It's been a long time since these jobs were ran in I'll disable them for now since they're not on the critical path. The integration tests for these builds are already disabled. Next step is to get the nix-build step working, right now it's failing to clone the |
|
Ok, i'm putting this on hold for now. The current problem with Priority now is to work on the agent migration and complete that, including removing |
|
@tegefaulkes is it possible to pull in the latest PK release right now? |
|
You should be able to use 1.1.5-alpha.0. |
|
Looks like |
|
To address 3., it's important to reactivate some integration tests. But they need to be written minimally, so that they don't rely too much on implementation details. We're all the way up to the UX now. Not in the weeds. @tegefaulkes @amydevs |
|
Docker image has been built and deployed, so that is all still working. However there are integration tests failing https://gitlab.com/MatrixAI/open-source/Polykey-CLI/-/pipelines/1040057667. This was expected, it's time to reactivate and refactor our integration tests, and maybe get PK re-enabled on windows and mac too. |
|
The We saw this before. I think the script used to load the quic native binding doesn't seem to work after being esbuilt or something. Not sure need to check. |
|
The error comes from: throw new Error(
`Failed requiring possible native bindings: ${prebuildTargets.concat(
npmTargets,
)}`,
);Which is in /**
* Try require on all prebuild targets first, then
* try require on all npm targets second.
*/
function requireBinding(targets: Array<string>): Quiche {
const prebuildTargets = targets.map((target) =>
path.join(prebuildPath, `quic-${target}.node`),
);
for (const prebuildTarget of prebuildTargets) {
try {
return require(prebuildTarget);
} catch (e) {
if (e.code !== 'MODULE_NOT_FOUND') throw e;
}
}
const npmTargets = targets.map((target) => `@matrixai/quic-${target}`);
for (const npmTarget of npmTargets) {
try {
return require(npmTarget);
} catch (e) {
if (e.code !== 'MODULE_NOT_FOUND') throw e;
}
}
throw new Error(
`Failed requiring possible native bindings: ${prebuildTargets.concat(
npmTargets,
)}`,
);
}The error message tells us that it is trying load native binding in: The first path is The second path is The reason is that:
|
|
I wonder if It could be due to It's possible that originally node-gyp-build does it differently, and so that's why it doesn't have this problem. |
|
I think I have an idea, it's possible the entire |
…ns the `package.json` is required to ensure that `main` can be resolved
|
Yep that was the problem and was solved by adding As a general matter it is somewhat more correct to depend on the special packages since the However it is little inefficient to bundle the One way is to change our custom However with the ESM migration MatrixAI/TypeScript-Demo-Lib#32, node's The docs suggest using https://nodejs.org/api/process.html#processdlopenmodule-filename-flags instead. So I think that would be more accurate, that means dropping the |
|
Ok remaining thing is:
|
|
If that passes then:
Then this should merge to master. And then we can focus on the remaining tasks in testnet 6. |
1. Making sure the docker integration tests pass in the relevant env parameters 2. extended timeouts for problimatic tests that do a password hash. Even though i'm sure we're passing in these options, the hashing still seems to be very slow. On the order of 10+ seconds. 3. Some tests were being skipped with `.skip` if this test was optionally disable based on the integration test it became `.skip.skip`. I've fixed this.
|
Pipeline Succeeded on 1040506440 for e029c8d |
4 participants
This is an automatic PR generated by the pipeline CI/CD. This will be automatically fast-forward merged if successful.