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
Even once Ouroboros Genesis is released, "disasters" or "inexplicably lucky" attackers may give rise to scenarios where a lightweight checkpointing mechanism would be useful. Notably, the mere existence of such a mechanism might even itself make the attack vector less attractive.
This Issue is to design such a mechanism.
At this point, the rough idea is that some configuration file could contain a list of points (ie pairs of a slot and a hash) that overrides normal chain comparison and/or validation in some way. The fundamental candidates are: EITHER a header/block whose point is in the list is considered invalid by fiat OR all headers/blocks in the same slot as one of the points but with a different hash are considered invalid by fiat.
This is explicitly not a heavyweight checkpointing mechanism, eg once that would give rise to a new and distinct genesis block.
The text was updated successfully, but these errors were encountered:
So far I've received feedback wondering if Mithril supplants this. I look forward to learning more about Mithril during the Consensus Office hours next week (2023 Oct 12), after which I can hopefully compare and contrast them here.
Edit: my initial guess is that Mithril would supplant it, but that creating Mithril snapshots is more involved than announcing a list of points. However, even "announcing" a list of points would require a process for signing them, so a significant portion of the overall Mithril process' "weight" is probably necessary for the lightweight mechanism of this Issue to actually be utilized by IOG/EMURGO/CF/etc.
Even once Ouroboros Genesis is released, "disasters" or "inexplicably lucky" attackers may give rise to scenarios where a lightweight checkpointing mechanism would be useful. Notably, the mere existence of such a mechanism might even itself make the attack vector less attractive.
This Issue is to design such a mechanism.
At this point, the rough idea is that some configuration file could contain a list of points (ie pairs of a slot and a hash) that overrides normal chain comparison and/or validation in some way. The fundamental candidates are: EITHER a header/block whose point is in the list is considered invalid by fiat OR all headers/blocks in the same slot as one of the points but with a different hash are considered invalid by fiat.
This is explicitly not a heavyweight checkpointing mechanism, eg once that would give rise to a new and distinct genesis block.
The text was updated successfully, but these errors were encountered: