twiggler

> That doesn't change how a Plugin is supposed to report what files it requires. Correct, because that is handled by `Plugin.get_all_paths()` I don´t feel particularly strong about `_get_paths` or...

> I wonder if this will really be any faster due to the extra overhead in parsing. We've optimized the compiler quite a bit already, with optimized struct.unpack calls. With...

For example, in `sqllite3.py` there are `Pages`, which come after the `header`. This fact is not obvious from looking at the cstruct spec, but I inferred it from `SqlLite3::raw_page`. With...

> If the goal is to be clearer about the file format, I believe the documentation should be improved. The goal is indeed to document our knowledge of file formats....

> Luckily most file formats we implement don't change that often 😉. True, but the code that parses them does (assuming that the documentation also references the python code). >...

> For example in SQLite, how are you going to implement WAL? In our current setup we can just return a different byte buffer for the requested page, how will...

> How would you implement monadic comprehensions? I have some intuition around coroutines Turned out this was already discovered independently by others. For example, see https://old.arrow-kt.io/docs/patterns/monad_comprehensions/, and https://parsy.readthedocs.io/en/latest/tutorial.html#using-previous-values The idea...

I think `serde` operates at a different level of abstraction, where Rust data structures are deserialized from a data format such as JSON or Bincode. Essentially, the data is already...

Hi @winingerori, I will provide you with an elaborate answer later this week

Interestingly enough, we are discussing to introduce Parsers as a concept to Dissect. Plugins can then delegate regex scanning (or more complicated parsers) to this Parser. The Parser has no...