Based on latest upstream

• A lot of test cases from many purescript libraries passed, special thanks to @Hong-Xiang
  only JS specific ones(like converting builtin type introspections) failed.
• change license to MIT #20
• update codebase to PureScript v0.13.8 #22
• fix namespace issues of getitem_looper and getattr_looper

allow large scale codegen & Python read IR speed up & bugfix

• codegen support for JavaScript-specific behaviors: add default value None for each function argument. fix #10
• faster codegen: avoid pretty print when not specified. #13
• lighter volume: for generated files. type level computing can produce huge amount of code, to make generated code Git maintainable, we use bzip2 to compress stuffs. #13
• options for pspy-blueprint changed(--py-dir <dir0> --entry-mod <A.B.C> --ffi-dep <xxx> --out-format [Pretty|Compact|Compressed]).
• special optimization for repeating indexing and attribute accessing.
• the topdown file format to allow Python to load large scale code, and load code faster. fix #13
• tagless final approach to support multiple data formats of output IR for codegen,
  check Topdown/{Raw.hs, Topdown.hs, Pretty.hs}..
• add CI tests based on a purescript-python project. #15

v0.1.1

2020-02-24: Version 0.1.1.0

Thanks to @glanham-jr for building the release of macosx-10.7.
Thanks to @lfkdsk for building the release of macosx-10.15.

• Fix multi-param datatype constructors

• Fix throwing exceptions

• Fix the use of some JS specific features in CodeGen:

  PS codegen uses JS tricks like "a" + ["b", "c"] == "ab,c",
  we change the generated code like throw new Error("a" + ["b", "c"])
  to throw Error("a" + ",".join(["b", "c"])).

  *Further optimizations can be done later.

v0.1.0.0

• License changed to LGPL for legal distribution of statically linked bianries.
• Remove some redundant code for making executable, implement the correct interface of MonadReader for codegen monad

preview release

0.1.0.0a

fix link