Our current error reporting loses information, especially when run against a workspace, as there are multiple errors with different stack traces. Currently we combine all the messages and re-throw a single error, which is unfortunate.
This new approach does something similar, but also persists stderr and stdout if it's been defined (as the error message is always useless coming from execa). It looks like the following.
Our current error reporting loses information, especially when run against a workspace, as there are multiple errors with different stack traces. Currently we combine all the messages and re-throw a single error, which is unfortunate.
This new approach does something similar, but also persists
stderrandstdoutif it's been defined (as the error message is always useless coming from execa). It looks like the following.Fixes #71