Improve user experience with better manifest errors messages#1229
Conversation
This is patch introduces suggestions in various error messages from the manifest parser. A potential suggestion is made using edit distance computing from a list of possible values. If the quality is not deemed good enough, the full list of values is printed.
|
Very cool! |
|
@mosteo somethings wrong with the macOS CI. I don't think they updated the |
|
I see... Strange, and different in each macOS setup. No immediate ideas to offer :/ |
|
@simonjwright we have some issues on macOS that I don't understand, in case you have some time to give it a look. |
|
I can find the tests, and I can run them locally, and I see failures. What I can’t see is how to run an individual test in diagnostic mode (i.e. so I can see what went wrong in detail). |
|
So |
|
Here are the tests which failed when run together from the top level. publish__check-pre-release-versionPASS publish__local-repoFAIL misc__did-you-meanPASS publish__local-repo-nonstdFAIL index__branch-mismatchFAIL misc__local-reject-originPASS build_profile__defaultPASS |
|
Looks like you don't have the same errors as we have on the GitHub CI. |
|
My trouble seems to be that I’m running git version 2.37.0 (Apple Git-136) which defaults the initial branch to |
Ohhh well spotted. |
|
I’m getting a PR together. Running CI in my repo now. |
|
Gosh it’s slow! the macos runner took 1h7m to install the toolchain, and has already spent an hour in Run test script (without Homebrew) and is still compiling |
|
I spoke too soon. See PR #1244. |
|
So by merging this we are going temporarily with the old macOS version, right? At least it buys us some time. |
This is patch introduces suggestions in various error messages from the manifest parser. A potential suggestion is made using edit distance computing from a list of possible values. If the quality is not deemed good enough, the full list of values is printed.