Tendermint: why use the mock.Mempool{} to initialize the blockExec

[banishee]

Tendermint version (use tendermint version or git rev-parse --verify HEAD if installed from source): 569981325a43c18e65beb193adf7d172a498a5bc

What happened:

• consensus\replay.go line 476 : blockExec := sm.NewBlockExecutor(h.stateDB, h.logger, proxyApp, mock.Mempool{}, sm.MockEvidencePool{})
• Why use the mock.Mempool{} to initialize the blockExec? I notice that the comment showing that Mempool is used for test
• if I add some functions in the interface Mempool, the structure CListMempool is easy to achieve the additional functions, but the structure Mempool is a empty struct. So it is hard for me to achieve the logic code.

What you expected to happen:

• use the CListMempool to initialize the blockExec

[zhongqiuwood]

It's weird that test code was used in production code.

[melekes]

It's simple. Because we don't need both of them (mempool and evidence pool) during replay (replaying missing blocks to application). mempool - we don't need to reap transactions. evidence pool - we don't need to reap evidence. block is already there (created before), Tendermint just replaying it to the application.

[melekes]

I hope https://github.com/tendermint/tendermint/pull/4785 clarifies things a bit. Thanks for bringing this to our attention!