Hey @Ronanb Browne
At the moment, the whole payload is sent at the end. In case of an interruption, nothing is sent. We can’t really send the data already captured “as-is” because it might be corrupted (missing important events that are captured at the end of the build, for instance).
In order to support this, we’d have to deal with sending data while the build is executing, and solve the problem of the required events from the end of the build.
r
Ronanb Browne
12/09/2022, 10:46 AM
Thanks for the comment , thank makes sense
f
François Guillot
12/09/2022, 10:47 AM
You’re welcome to push it up the chain with your customer success representative!
✅ 1
c
Claudio Roberto França Pereira
01/05/2023, 8:27 AM
Perhaps sending as results come in is too big a change to be reasonable, but what about handling the kill signal, sanitizing what is available and uploading it as a partial build scan? Maybe that's more feasible?
Is there a ticket about partial build scans being uploaded?
f
François Guillot
01/05/2023, 9:50 AM
The sanitization process might not be that easy. We have lots of events and ids to cross-reference events so we somehow need to make sure all of that is perfectly sanitized. Definitely doable but not trivial. Or we sanitize it to the absolute minimum of events (eg no tasks, no test, ...).
I’ll relay this request internally but I can’t give you an ETA nor if it will be done at this point