Open baybal opened 3 years ago
would find this also really useful. I have to disable icecc in some yocto recipes due cc errors. Also I'm curious (I don't think so) if it distinguishes between some failure on the "client" contacting the remote iceccd (like remote has "disappeared" during an operation) vs a compile error. I had some failures also when some decent amount of nodes went away due network changes.
I believe the retry mechanism could differentiate on how many retries should be done remotely before a local fallback (if at all) should be done before the error being reported.
As said in topic, it makes sense to retry failed commands locally to see if their failure was due to code itself, or the compiler.