Closed rootulp closed 1 week ago
The alternative approach would be what @cmwaters originally suggested which is to remove authz.MsgExec
from the list of acceptedMsgs.
but my concern with adding this is that even though a MsgExec inside a MsgExec seems like a strange use case, what if it's actually desirable for certain scenarios.
I think it's sufficient to leave it as is (with the assumption that there is no valid use case) and only make an adjustment if a user presents a valid use case to modify it
Just to clarify are you in favor of:
Option A. allowing nested authz messages (the current implementation) Option B. enforcing some upper limit on recursion for authz messages (a.k.a something like https://github.com/celestiaorg/celestia-app/pull/3590 or split out into it's own ante handler).
Option A.
Sorry for the confusion. I actually changed my mind on this. I think if we want to prevent nesting it should be it's own antehandler. For now, we should leave this ante handler to allow nesting
Thanks for clarifying. Do you think adding a new ante handler should block v2?
Closing this as won't do and created a new issue b/c we don't have clarity yet: https://github.com/celestiaorg/celestia-app/issues/3609
Do you think adding a new ante handler should block v2?
Personally, no. And as has been discussed, it should probably be a CIP
An alternative option to the one Callum proposed is:
but my concern with adding this is that even though a MsgExec inside a MsgExec seems like a strange use case, what if it's actually desirable for certain scenarios.
I don't see a major risk from not adding this constraint because if a user crafts a deeply nested MsgExec wrapper (i.e.
MsgExec(MsgExec(MsgExec(...)))
then the transaction will be rejected based on tx size._Originally posted by @rootulp in https://github.com/celestiaorg/celestia-app/pull/3555#discussion_r1636764149_