I'm making a Spin compiler. Why?
-
Spin is an interesting language for its limitations
-
LameStation has years of content built around a subset of Spin
-
The ecosystem is fragmented and not fully specified
-
A documented lexer/parser would enable many kinds of analyses
-
This gives me an excuse to learn Go
Notes:
lame COMMAND [OPTIONS]
-
To compile Spin objects:
lame build
-
To load Spin objects over serial:
lame load
-
To enable a live build/load/debug cycle over serial
lame dev
-
To reformat Spin code accurately:
lame fmt
-
To generate documentation from any Spin object:
lame doc
-
To push and pull Spin objects to/from a remote repo:
# upload an object to the repo lame up # pull an object lame down pst=1.0.0
-
To support multiple code generator backends (Propeller VM?)
-
To write a (semi-)formal specification for the Spin language
-
To perform code generation for the Propeller 1
-
To support an alternative VM target for desktop / browser use
-
To support direct addressing via the
@
symbol -
To add a path system to Spin:
OBJ obj : "path.to.module"
-
To add automatic constant folding
-
To reduce the number of operators / keywords
-
Identity - Translate the input to the AST and re-emit to the output
-
Reverse engineer Propeller binary format