ProviderContextManager is not concurrency-aware

[fubuloubu]

Ran into this with a Silverback bot using the new FORK_MODE PR in the Silverback SDK (currently solving with an async concurrency lock)

Not sure this can be resolved without full async support of Ape's network management stack

[linear]

APE-1848 `ProviderContextManager` is not concurrency-aware

[fubuloubu]

Not quite sure what the issue was, but it lead to weird "trying to fork network mainnet-fork is not possible" or "ecosystem 'base' has no network mainnet-fork-fork" bugs

[antazoey]

Not quite sure what the issue was, but it lead to weird "trying to fork network mainnet-fork is not possible" or "ecosystem 'base' has no network mainnet-fork-fork" bugs

i can at least improve the error

[antazoey]

Maybe this helps a little? #2349

[fubuloubu]

Maybe this helps a little? #2349

Yes I believe that will prevent the error, but not totally ideal

I will learn if it works well enough or not when I try to implement backtesting to silverback (at some later date)

for reference, the backtesting mode of silverback takes a range of blocks and executes each block (and relevant event logs/transactions) in an isolated fork that only exists for the span of that one block

[antazoey]

Wha about a fork-lock for each network so you can't fork until you hold the lock, waiting for other threads / processes to release the lock

[fubuloubu]

Wha about a fork-lock for each network so you can't fork until you hold the lock, waiting for other threads / processes to release the lock

I don't think we need that now with the synchronous version, I will test this fix to see if it works good enough for my use case and get back to you