You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Nov 15, 2020. It is now read-only.
Now that networking is getting ready for release, we have client-side bot runners. This allows us to capture bot stderr for local logs.
The question that remains is how we will actually handle this data. The issue that arises is that our log is event-based, and stderr is just a data stream. One approach would be to treat a line on stderr as an event, and log it as it occurs. Apart from the mild inconvenience that a three-line error trace would be three events in this case, this might cause buffering issues (some buffers might only get flushed after enough data is written, some might be flushed once the process exits). Since the bot writer does not always control this buffering (for example, error output by the python program when writing a python bot), we might have an issue there.
All input on designing a robust mechanism would be welcome :)
The text was updated successfully, but these errors were encountered:
Now that networking is getting ready for release, we have client-side bot runners. This allows us to capture bot stderr for local logs.
The question that remains is how we will actually handle this data. The issue that arises is that our log is event-based, and stderr is just a data stream. One approach would be to treat a line on stderr as an event, and log it as it occurs. Apart from the mild inconvenience that a three-line error trace would be three events in this case, this might cause buffering issues (some buffers might only get flushed after enough data is written, some might be flushed once the process exits). Since the bot writer does not always control this buffering (for example, error output by the
python
program when writing a python bot), we might have an issue there.All input on designing a robust mechanism would be welcome :)
The text was updated successfully, but these errors were encountered: