Sam Rushing

The first thing people do when bootstrapping is run genffi (in the VM) to get `posix_ffi.scm`. It'd leave a nicer first impression if the earley parser was sped up. [not...

naming cleanup (pxll, BOX_INTEGER, etc...)

1

There are several naming issues that have accumulated over the years. * s/pxll/irken * s/BOX_INTEGER/TAG_INT * lib/pair.scm -> lib/list.scm * LIST/PUSH -> list/push These need to be done in one...

type errors with 'object protocol'.

1

The type solver is not catching some type errors when using object support (i.e., `ob::method` syntax). This particular case involved iterating over a map, I accidentally swapped the types of...

toplevel out-of-order objects.

1

Segfault. Looks like some extra care is needed with toplevel objects. Repro: ``` scheme (define some-object (object-maker)) ;;... use some-object ... (include "object.scm") ``` Need to look at `reorder-fix` and/or...

Currently, letreg is somewhat broken with the llvm backend, because assignment cannot be implemented due to SSA. If we map letreg variables to the `alloca` insn, then we can handle...

For example: `#\u0a == #\newline`.

Records have a tag attached for disambiguation. There is currently nothing done when the number of unique records exceeds the tag space available. 1) we need to throw an error...

record patterns ignore the closed/open distinction.

1

Because of the way record patterns are compiled, the distinction between a closed and open record is lost. All records patterns are treated as open. This: ```scheme (define thing1 {a=1...

matching two pvariants with accidentally different sub-types

2

This appears to be allowed when it should not. ``` (:thing x (specific:cons) ...) (:thing x (other:cons) ...) ```

type annotation lost when function inlined

2

see tests/t_anno3.scm