Nick Terrell

I’ll put up a PR that fixes the bug, so to_ssa.py generates correct code. Then I’ll think about a different pruning approach.

I think the allowance of undefined variables in `phi` nodes may have been a mistake. The wrinkle that `phi` nodes are allowed undefined values, but no other instruction is, really...

I've added support for the memory extension to `to_ssa.py`. In order to do that, I needed to support `null` pointers in `brili` and `brilirs`. Null pointers are created with `alloc...

Isn't your `defined_analysis()` an over-estimate of the defined variables, so it will miss some cases? I believe the problem you're trying to solve: ``` a.1: int = phi a.0 a.0...

I've put up PR #119. I believe that it fixes all the issues of #108.

> can you explain what you mean by "over-estimate of defined variables"? Consider the variable `v` in the `loop` function in this example. I believe your analysis will say that...

There is no reason it couldn't be open sourced, its only dependency is xxhash. It currently only provides an interface to the `XXH64()` function, but it could be extended if...

I would be okay accepting a PR that enabled OpenMP support for divsufsort, as long as it is hidden behind a define that defaults to disabled. We can then thread...

For CPU, you'll need to use an external tool, but unless you explicitly enable multi-threading, each compression is single-threaded. For memory, we provide `sizeof` functions for each of our objects,...

Thanks for the report! We'll fix this bug in the next release. We've had other bugs in the decoder that rejected valid frames. I will put up a PR shortly...