Update Code to 1.81.1 #6385
Update Code to 1.81.1 #6385
Conversation
|
Increase of build timeout is not required. It took something over 30 mins on my machine. ℹ️ The pipeline for #6369 failed due to an e2e test error. |
|
Functionality [modified by patches] tested and found to work:
Jupyter Notebooks also work fine:
|
|
Weird, I am not sure why the e2e tests are failing. I re-ran them a few times but it seems like code-server is not loading in time, and I am seeing similar issues locally, but the really weird thing is that it only happens if I have the browser dev tools open. Maybe they added some kind of debug tool and it runs with dev tools open or in the e2e tests for some reason and is hanging. I will be able to investigate more next week. |
Thank you. It would be great to have a release by the end of this week. |
|
Ping @code-asher |
|
I have been trying to reproduce the hang but I cannot seem to get it to happen again. I updated the callback path to account for the base path. In addition to the tests above, I also tested display languages and the GitHub token injection (since some code around secrets changed). I get a lot (~23) of "throttler is disposed" errors when closing the browser but it does not seem to cause any issues so I think we can ignore it. There is an open issue for it: microsoft/vscode#188929 Assuming CI passes I think we will be good to go. |
|
e2e tests still timing out unfortunately. |
I think the test failures are actually because Node modules are not being installed so the page never loads, and for some reason npm install is not exiting with non-zero when that happens.
pipefail would be ideal here but not sure how wide the support is yet.
code-asher
force-pushed
the
code-1.81.1
branch
from
1d39e89
to
99837dd
Compare
|
Sorted the test hang, but now there seems to be some syntax issue preventing WebKit browsers from working. |
|
I tested by inserting "ls -alh /home/runner/work/code-server/code-server/lib/vscode/" into the "test-e2e.sh" file. And the image below is the result of adding "with" to "actions/checkout@v3" in "build.yaml". Regrettably, this isn't a solution. An error occurs that's very similar to what you all are experiencing. Looking at the git history, not including the value "submodules: true" when calling "checkout@v3" has been done for a long time, so I don't think it's due to that. However, since the two results are different, I hope it serves as a clue to solve the problem. I sincerely hope the issue will be resolved soon. |
|
Thank you for experimenting with it. I ran the CI build in Safari and the syntax error seems to be coming from some They are coming from here: It does not seem like this file has changed recently, so I wonder if the compilation target was changed and now it is outputting JavaScript that is not quite compatible with Safari. |
|
Oh, to clarify the submodule, the e2e step downloads the built package from CI and uses that for the tests, so there is no need to populate lib/vscode. |
|
In 1.81.1 static {
this.white = new I(new E(255, 255, 255, 1))
}In the previous version we got: I.white = new I(new E(255, 255, 255, 1))I get the same error on vscode.dev so it appears to be an upstream issue. |
No worries. We can also skip this version. If you feel the same, please close this I only prefer versions that work absolutely fine for all browsers. |
|
Let us focus on the next release, then: microsoft/vscode#189778 |
|
Sorry all, I was out last week so I was not able to get back to this. I think skipping makes a lot of sense, we will need to see if we can get it patched upstream in the meantime if it has not already been. |
Ping @code-asher