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
For debugging it's useful to have tests deterministic (typically when some hashes are messed up or you want to compare your broken branch with master). We currently don't support that and I think we could achieve it by modifyingFr.random(), GrumpkinScalar.random() functions such that they use a central source of "randomness" and this source could be a pseudorandom generator with some seed. If a SEED env variable would be set this pseudorandom generator would be used with the given seed instead of the truly random numbers.
The text was updated successfully, but these errors were encountered:
Fixes#3899
When exporting SEED=[number] env var deterministic mode gets used now.
Note that there are other sources of info in some tests so the results
might not be the same in all the cases. E.g. in integration l1 publisher
test the second block has a populated timestamp and therefore will have
different archive root. For first block the timestamp is set to 0 so
there an equal result should be achieved.
For debugging it's useful to have tests deterministic (typically when some hashes are messed up or you want to compare your broken branch with master). We currently don't support that and I think we could achieve it by modifying
Fr.random()
,GrumpkinScalar.random()
functions such that they use a central source of "randomness" and this source could be a pseudorandom generator with some seed. If a SEED env variable would be set this pseudorandom generator would be used with the given seed instead of the truly random numbers.The text was updated successfully, but these errors were encountered: