todo.md

• optimize
  • tail opt
  • remove types from expr
  • rc all values
  • rc all types?
• change member access
• add methods
• struct
• generic native func
• generic user func
• generic structs
• recursion
• generic native type
• arrays
• bools
• rationals
• strings
• spliced arrays
• sets
• queues
• inner functions
• function types
• return inner function?
• fix closures, make the scope chain RC'd (or ARC'd) and find a smarter way to merge the closures
• optional
• cast (some(1) || error(''))
• operator overloading
• default params
• default params in native functions
• generic specialization (with UD, i think there's a bug)
• less copies
• more tail call optimizations? ([] + ... => ...)
• avoid specializations for unknown types
• limits
  • depth limit
  • recursion limit
  • size limit
• unions
• recursive unions
• make struct binds use vectors
• lambda functions
• blank structs/unions
• dynamic native functions (Array<T> == Array<T>)
• tuples
• tuple as recursive member
• mappings
• double quoted strings
• bundle the root compilation scope and the interner to one struct
• I think we turn static functions to real function all the time instead of once
• root evaluation scope
• delete every function, pub, and derive we can get away with
• replace all vec params with slices
• make it so we don't have to clone declarations
• integrate rtcell into the root scopes
• small ints shouldn't need heap allocation
• make defaults evaluate at runtime
• can assert detect when equality happens and report it?
• make turbofish useful
• more std
  • sets
  • sorting
  • dict delete
  • dict pop
  • hash methods
  • cmp methods
  • aggregators
  • range
  • filter
  • any, all
  • takewhile, skipwhile
  • take, skip
  • multizip
  • unzip
  • enumerate
  • assert
  • array multiplication
  • str concat
  • str multiplication
  • str split
  • str chars
  • regex?
  • binary search/insort
  • partial
  • nth
  • first/last
  • default equals/hash/cmp/to_string for many types
  • auto mapping for types with eq/hash
  • auto sort
  • infinite sequence
  • slice sequence type
  • ne for most native types
  • member_wise
  • is_error/if_error with error matching
  • get_error
  • dynamic "display"
  • add str to something that implements to_str?
  • is_close
  • bool xor
  • float pow
  • sqrt
  • nth root
  • nth last
  • aggregators
  • reduce should be in generators
  • everything in excel
  • atan2
  • geometric distribution
  • some things in python
  • sequence::manifest
  • json?
  • chain sequence?
  • set xor
  • xors
  • sleep
  • chunks
  • random
  • is_infinite
  • mean/std for distributions
  • sort reverse (dyn)
  • matrices
    • eye
    • full
    • diag
    • determinant
    • inverse
    • transpose
    • multiply
    • map
  • complex numbers
  • time of day?
  • list shuffle
  • numerically safe addition
  • byte types?
  • better flatten
  • complex trig?
  • lru_cache?
  • compression/encryption/hashing
  • generatoe/seq count(i, eq)
  • threading??
  • json of option
  • base64
  • bit_or and bit_and for bools?
  • sinh
  • fix the mean/geo-mean implementations to be better and clearer
  • counter_mode generic
  • xray version const
  • to_hash key adapter
  • optional::map_or should have 2 generic params
  • most dyn functions should have non-dyn versions
  • a lot of the sequence/stack functions can be removed
  • generator::all/any (bools)
  • delete sequence::group
  • sequence::is_infinite
  • improve rank_avg and rank_eq to be more efficient and generic
  • stack::to_array and to_array_reversed should be flipped
  • sub{date, dur}
  • make sample_variance/sample_standard_deviation generic
  • duration cmp/eq/format?
  • fraction constructor uses float div and there's no reason for it
  • int trunc_div
  • split debug to a function that gets the internal rep, and one that prints it
• tests for:
  • is_error
  • if error (inc. tail recursion)
  • display
  • bool display
  • display from UD type
  • bool hash
  • bool cmp
  • float mod
  • float div
  • float ceil
  • float trunc
  • float cmp
  • float to str
  • int bitwise ops
  • int pow
  • int ne
  • int hash
  • int cmp
  • sequence rpush
  • sequence insert
  • sequence set
  • sequence swap
  • str hash
  • str cmp
  • function call in default (inner function call in default?)
  • internal structs
  • test all kinds of failures
    • including runtime violations
  • get operator
  • mapping get
  • hexa/bin numbers
  • scientific notation
  • float neg
  • is_close
  • float sqrt
  • float pow
  • mapping entries
  • all the integer edge-cases
  • zip1
  • unzip1
  • repeat (with n and infinite)
  • lambda with defaults
  • str find
  • str split/nsplit/rsplit edge cases
  • members for recursive types
  • errors:
    • VariableAsType
    • OverloadAsType
  • sin
  • turbofish with zero params
  • dyn with with zero params
  • sum/product/mean/geomean? with complex types
  • str reverse
  • negative datetimes
  • duration operations
  • negative duration
  • date sub/add
  • date/datetime str
  • format_replace
  • flatten with empty
  • map_values
  • mapping::clear
  • str::contains
  • int::digits
  • windows that are wider than the original generator
  • to_stack
  • add{date, duration}
  • negative durations
• reduce macros in builtin/core
• better errors
  • compilation error tracing
  • dynamic binding error tracing
  • runtime error tracing
• get operator
• auto type for specialization (foo::<., Stack<int>>(1, stack()))
• cells instead of dicts for namespaces
• All the TODOs
• all the warnings
• test scripts
• imports/include files
• turn this to a library
• documentation
• ! operator (as shorthand for .value)?
• code documentation
• newline/escape chars in literals
• straighten out the "errors in container/compound" question
• find and remove duplicates
• string interpolation?
• auto return type (fn foo () -> $)
• in native functions, when you're going to return something big (like when adding two sequences), check that you have room for it first
• type aliases?
• total function call limit
• maximum search limit (count().nth(0, (x: int)->{x<0}))
• execution time limit
• "side effect" limitations (allow/forbid sleep, display, assert)
• declarations in lambdas
• cache xtypes?
• create a native type registry, so we don't discern them by name
• cache factory functions
• forward references?
• native dynamic functions?
• line number in RTCELL?
• warnings (like when assigning variable to magic method?)
• make [str].sum use join
• custom seeding for hash
• make sure nan inf and negzero are impossible
• time provider?
• managedXError should be a cow
• upgrade pest
• WASM