Benjamin Bannier
Benjamin Bannier
In order to detect whether to send out a failure email we previously would check that the action conclusion contained certain statuses. This check seems to have been nonsensical as...
When parsing of a field fails we currently select the next `&synchronize` field _after the failing field_. This means that if parsing fails on the last field of the unit...
In #1398 we discussed an optimizer pass to make unaccessed named field anonymous. We should have a look at how hard this would be to add, maybe even without adding...
If a function parameter is never read we could add an optimizer pass which either completely removes them (requires edits and both the function declaration and the users), or at...
We currently generate temporaries which depending on user code might be dead. We should look into an optimization to remove them. E.g., for the following unit ```ruby public type X...
Currently `Chain` is a thin wrapper around `Chunk` which provides singly linked list functionality. The current implementation makes it harder than needed to e.g., implement an efficient `findChunk` on top...
If one imports another module we always emit code for all `public` units in the imported module, even if none of them were used. This makes it hard to move...
### Check for existing issues - [X] Completed ### Describe the feature I am working with a tree-sitter grammar which separates the tree-sitter generated `src/` directory into a separate Git...
Currently `kondo` [hardcodes a list of possible CMake build directories](https://github.com/tbillington/kondo/blob/415cd335efeee3f251513eb6b9db3a8f8dd90a69/kondo-lib/src/lib.rs#L46). This often works well enough, but since the name of a CMake build directory is merely convention it can fail...