Replies: 5 comments 3 replies
-
|
This is interesting... I can understand that when the agent runs the tests itself (without the extension), it can easily parse the terminal output to get the test result. The question is, how does it know how to work with external extensions? I wondered if you could tell it which view or terminal it should look for test results, maybe then it can correctly parse the result output? 🤔 |
Beta Was this translation helpful? Give feedback.
-
|
I can try to play with it a bit more and see if I can somehow tell it to look in another place. I assumed that this extension integrates with VSCode testing features (e.g. Test Explorer, Test Coverage) and the agent somehow hooks into this as well. Also, it would be awesome if the agent was able to run unit tests with test coverage and you could ask it to write tests only for the parts without coverage. |
Beta Was this translation helpful? Give feedback.
-
Indeed, it does.
You would hope...
Yeah, but what are human developers going to do then? 😅 |
Beta Was this translation helpful? Give feedback.
-
We're all asking the same question 😅 |
Beta Was this translation helpful? Give feedback.
-
|
@connor4312 might be able to help In fact the best might be if you open this issue on the vscode side https://github.com/microsoft/vscode-copilot-release and ping me @isidorn - since it looks like Copilot is wrongly picking up test results (and might not be related to jest) |
Beta Was this translation helpful? Give feedback.
-
|
Thanks! Done here microsoft/vscode-copilot-release#7740. |
Beta Was this translation helpful? Give feedback.
-
|
Looks like it's caused by the fact that:
According to microsoft/vscode-copilot-release#7740 (comment) it looks like Copilot might be able to handle this soon, if I understood this comment correctly. |
Beta Was this translation helpful? Give feedback.
-
|
According to @connor4312
If this means to look at only the first |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hey!
I'm trying out the new Copilot Agent on a simple React project created with
create-react-app. It provides thetestcommand in thepackage.jsonthat runs all unit tests.With this extension enabled, when the agent makes some changes and runs unit tests to verify them, it will always say that all tests passed successfully, even if they haven't.
It will run a command like:
and say something like:
When I disable this extension, it will ask me if it can run
npm testcommand. It has to be modified tonpm test -- --watchAll=false, as thetestcommand provided byreact-scriptsruns in the watch mode, but now when tests fail, the agent will correctly notice it and try to fix them.Any idea why it might be happening? Could it be something caused by the reporter setting? Does anyone else have the same issue?
Beta Was this translation helpful? Give feedback.
All reactions