Bernhard Mueller

I tried adding a hack in `symbolic.py`, didn't work (laser still returns 0 open states).  It works with `--max-depth 64` on the cmdline...

Hi @gkevinyen5418, the documentation is outdated. AFAIK, if a Solidity file contains multiple contracts, Mythril should analyze the creation bytecode of each contract independently. In the case of callgraph generation...

What we did in [Sabre](https://github.com/b-mueller/sabre) is [take the contract with the largest bytecode](https://github.com/b-mueller/sabre/blob/db8db7fb260f635ec653a7f72da1ebb79b0df3cf/lib/compiler.js#L236) if no contact is explicitly specified. The reasoning was that if there's multiple bytecode outputs, the largest...

Yep, this would be an improvement. In the meantime you can use the following hack to determine what's being executed: ``` def is_prehook(): """Check if we are in prehook.""" return...

IMO the strategy should be to always be up-to-date with the current mainnet code by default. It would be cool though to be able to switch to an upcoming fork...

Bump :)

Interestingly this happens only with bytecode compiled with newer solc version. Octopus processes the same contract when compiled with 0.4.24 but fails with 0.5.7. ``` pragma solidity ^0.5.0; contract Test...

We're scrapping the old benchmarks so this can be closed.

What model are you working with and what does your `.env` look like? Try setting `MAX_CONTEXT_SIZE`, `MAX_MEMORY_ITEM_SIZE`, and `SUMMARIZER_CHUNK_SIZE` to lower values.

Try the values for gpt-3.5-turbo: ``` MAX_CONTEXT_SIZE=2000 MAX_MEMORY_ITEM_SIZE=1000 SUMMARIZER_CHUNK_SIZE=1500 ```