A data format that serves a similar purpose to JSON, but with the following features:
- Readable - less nesting, less noise
- Minimal and compactable
- Extendible with type tags
- More primitive types, plus multiline strings
Objects/maps/dictionaries are called tables here and are sets of key/vals
// Tables and nesting
table =
child1 = 1
child2 = 2
child3 =
child4 = 4
// Nested keys can be set from anywhere with a full path
table/child4/key = 'val'
// Inline can be done with semicolons and parens
inline_table = (x = 1; y = (z = 2))
Sequences are ordered collections of values
sequence =
# 1
# 2
# 3
inline_sequence = # 1 # 2 # 3
sequence_of_tables =
# x = 1
y = 2
z = 3
# a = 1
b = 2
inline_sequence_of_tables = # (x = 1; y = 2) # (a = 1; b = 2)
Entries in a table can be added at any point using a path with slashes. This helps to reduce overly nested configuration files:
beer =
name = "super hefeweizen"
category = "wheat beer"
beer/description = """
This is a very long, multi-line description.
I wouldn't want this block of text in a nested format above.
So instead I can set it down here.
"""
// Let's add some more details
beer/attributes =
bitterness = 0.1
shade = 0.22
sweet_dry = 0.88
rating = 89
Likewise, elements in a sequence can optionally be added at any point without nesting:
my_fav_numbers = # 1 # 2 # 3 # 5
my_fav_numbers/++ = 7
my_fav_numbers/++ = 11
// For appending elements, order in the document will naturally matter
You can tag values with $tag. This can be used as a type annotation:
my_uuid = $uuid "2a867683-f1b8-4b2e-be25-b7aa92e2e6c"
one_tenth = $ratio # 1 # 10
my_grade = $percent 91
fruit_names = $set # "apple" # "banana" # "grape"
my_customer = $customer (name = 'Bob Ross'; orders = 3)
// Tagging a table or sequence with linebreaks is easy:
my_order = $order
product = 1234
quantity = 2
profile_tags = $tags
# "admin"
# "editor"
# "scientist"
# "author"
users =
# $user
username = 'bobross'
ip_address = $ip # 128 # 3 # 66 # 210
login_info = $login_info
login_count = 10
last_login = 2018-10-19T23:11:11.244Z
Built-in primitives:
float1 = 1.0
float2 = -33e-10
int1 = 1
int2 = +99
int3 = -99
date_time = 1979-05-27T07:32:00-08:00
date = 1979-05-27
time = 09:09:11.1234565
time2 = 09:08:07
hex = 0xDEADBEEF
hex2 = 0xdeadbeef
oct = 0o755
binary = 0b11011010101
regex = /^\s+\w\n\s+\w\n\s+$/
bool1 = true
bool2 = false
single_quote_string = 'content'
double_quote_string = "content"
multiline_string = """
Hello
world
"""
All rdn structures can be collapsed into a single line without whitespace:
// starting with:
x =
y =
# a = 1
b = 2
# c = 3
p =
q = 3
x/z = 4
p/r = 5
// collapses into:
x=(y=(#a=1;b=2#c=3)z=4)p=(q=3;r=5)
Punctuation marks (/, =, ;, $, ++, #, ', "), when needed inside keywords, can be escaped by
either prefixing it with a backslash or surrounding the keyword in quotes