Sjors Provoost
Sjors Provoost
Concept ACK. But rather than relying on block invalidation, I would prefer to have a clean rollback. That would be useful as it's own RPC too: turn off network, call...
> > Seems like this should freeze indexes, wallets, etc, and hold a lock so nothing else happens during the process...? > > > But rather than relying on block...
> I am still planning to do the rollback in a cleaner way but[...] It would become a refactoring follow-up that does not change the RPC interface again. That makes...
re-utACK ac9d53d0dee26db58ab125aa369f476c232da660
re-utACK 46c8d75d24d3355ec468f7f608effe94636e16db #30690 just modified the script this PR deletes.
re-utACK 94b0adcc371540732453d70309c4083d4bd9cd6b Just a rebase and documentation improvement.
utACK cbcb2c82669deaad71e739c64b1baf687e76e604 Holding on to `cs_main` in this particular spot, when index catches up to the tip, is not a big deal performance wise. The other, much more frequently called...
@ryanofsky's suggestion looks good to me as well. Though maybe if the timing is really unlucky you'd get lots of `Commit()` calls?
I suppose the original idea was to run these first as a sanity check. To somewhat retain that property, perhaps the entire test suite can be aborted if this fails....
I was assuming that any failure of the framework tests means your system is broken, but maybe that's too strong an assumption.