nada
Use the nada
CLI tool to manage Nada projects. It can:
- Create a new Nada project
- Compile Nada programs
- Get program-requirements, the runtime requirements of the preprocessing elements of a program
- Generate test files for Nada programs
- Run Nada programs
- Test Nada programs
- Get benchmark summaries of Nada program tests to measure performance
Usage
Install the Nillion SDK to use the nada
tool
Usage: nada <COMMAND>
Commands:
init Create a new nada project
build Build a program
run Run a program using the inputs from a test case
test Run tests
benchmark Benchmark one or multiple programs
generate-test Generate a test for a program with example values
program-requirements Get requirements for program
shell-completions Generate shell completions
list-protocols List Protocols (instructions) in JSON format
publish Publish a nada program
help Print this message or the help of the given subcommand(s)
Create a new project
You can create a new Nada project by running
nada init <project-name>
This will create a new directory with the project name and the following structure:
nada-project.toml # File to configure the project
src/ # Directory to store the programs
src/main.py # nada program
target/ # Directory where nada will produce the compiled programs
tests/ # Directory to store the tests
nada-project.toml
file
This file defines the nada project, it looks like this
name = "<project name>" # name of the project
version = "<version>" # version of the project
authors = [""] # authors of the project
[[programs]] # program entry, you can have several of these
path = "src/main.py" # path to the program
prime_size = 128 # prime size to use in the execution of the program
nil-sdk.toml
file
We recommend creating a nil-sdk.toml
file in the project directory to specify the version of the Nada SDK to use.
The file should have the following format:
version = "<version>"
This will ensure that when you call nada or any other Nillion SDK command, it will use the version specified in that file.
Set up a virtual environment
-
Create a python virtual environment
python3 -m venv .venv
-
Activate the python virtual environment
source .venv/bin/activate
-
Install nada-dsl from pypi
pip install --upgrade nada-dsl
- Create a python virtual environment
python3 -m venv .venv
- Activate the python virtual environment
source .venv/bin/activate
- Install nada-dsl
pip install --upgrade nada-dsl
Add a new Nada program to the project
To add a new program to the project, you can create a new file in the src
directory and add a new entry in the nada-project.toml
file.
There is the option to use another name for your program using the name
field in the program entry.
Let's say that the new file is called new_program.py
and you want to call it my_program
, then you can add a new entry in the nada-project.toml
file like this:
[[programs]]
path = "src/new_program.py"
name = "my_program"
prime_size = 128
Build (compile) a program
To build all the programs in the project, you can run
nada build
Note: this will only build the programs specified in the nada-project.toml
file. Visit the section Add a new program to the project for more information.
Also you can specify which program to build by passing the program name as an argument
nada build <program-name>
The build
command will produce a <program name>.nada.bin
file in the target
directory.
You can use this file to upload and run it on the Nillion Network.
Get program requirements
The program-requirements
command prints the runtime requirements of the preprocessing elements of a specific program.
nada program-requirements <program-name>
Example program and program-requirements:
from nada_dsl import *
def nada_main():
party1 = Party(name="Party1")
party2 = Party(name="Party2")
party3 = Party(name="Party3")
a = SecretInteger(Input(name="A", party=party1))
b = SecretInteger(Input(name="B", party=party2))
result = a + b
return [Output(result, "my_output", party3)]
Here are the program requirements of the 3 Party addition program above:
Requirements:
ProgramRequirements {
alpha_elements: 2,
runtime_elements: {
Lambda: 2,
},
}
Generate a test file
To be able to run
or test
programs we need a test file with the inputs and expected outputs.
nada
has the ability to generate a test file with example values for you.
To generate a test file you can run
nada generate-test --test-name <test-name> <program-name>
It will generate a test file with the name <test-name>.yaml
in the tests
directory.
You should edit the test file to change the inputs and the expected outputs.
Run a program
To run
a program using nada you need a test file, visit the Generate a test file section to see how to generate it.
Run in normal mode
To execute a program you can run
nada run <test-name>
It will run the program associated with that test file and print the output. Similarly to the nada-run tool, it executes a Nada program using the Nada VM with the cryptographic algorithms but without the peer-to-peer (p2p) network.
Run in debug mode
nada
has the ability to run a program in debug mode.
In this mode, all operations are computed in plain text, so you can inspect the intermediary values.
When running in debug mode, you can view the operations performed in the program along with their corresponding values, as it does not execute the cryptographic algorithms.
To execute a program in debug mode you can run
nada run --debug <test-name>
This will run the program associated with that test file and print all operations, values and the output.
- Debug in plain text
- Nada program
- Test file
(.venv) ➜ nada-by-example git:(main) ✗ nada run --debug addition_test
Running program 'addition' with inputs from test case addition_test
Building ...
Running ...
Header:
Party: Alice id(0) # party_alice = Party(name="Alice") -> addition.py:4
Inputs:
iaddr(0) rty(SecretInteger) = Input(num_1) # num_1 = SecretInteger(Input(name="num_1", party=party_alice)) -> addition.py:7
Outputs:
Party: Bob id(1) # party_bob = Party(name="Bob") -> addition.py:5
Inputs:
iaddr(1) rty(SecretInteger) = Input(num_2) # num_2 = SecretInteger(Input(name="num_2", party=party_bob)) -> addition.py:8
Outputs:
Party: Charlie id(2) # party_charlie = Party(name="Charlie") -> addition.py:6
Inputs:
Outputs:
oaddr(0) rty(SecretInteger) = Output(sum) addr(2) # -> addition.py:10
Literals:
Loading Literals:
Loading Inputs:
Computing:
addr(0) rty(SecretInteger) = Load iaddr(0) # num_1 = SecretInteger(Input(name="num_1", party=party_alice)) -> addition.py:7
SecretInteger(30 mod 340282366920938463463374607429104828419)
addr(1) rty(SecretInteger) = Load iaddr(1) # num_2 = SecretInteger(Input(name="num_2", party=party_bob)) -> addition.py:8
SecretInteger(10 mod 340282366920938463463374607429104828419)
addr(2) rty(SecretInteger) = Addition addr(0) addr(1) # sum = num_1 + num_2 -> addition.py:9
SecretInteger(40 mod 340282366920938463463374607429104828419) = SecretInteger(30 mod 340282366920938463463374607429104828419) + SecretInteger(10 mod 340282366920938463463374607429104828419)
Loading Outputs:
oaddr(0) rty(SecretInteger) = Output(sum) addr(2) # -> addition.py:10
SecretInteger(NadaInt(40))
Program ran!
Outputs: {
"sum": SecretInteger(
NadaInt(
40,
),
),
}
loading...
loading...
Run and get metrics
Run a program with the --metrics flag to retrieve detailed execution metrics of your program. You can specify the format for the metrics output using the following values:
- text: Outputs metrics in a human-readable format directly to the console.
- json: Outputs metrics in JSON format and saves it to a metrics.json file.
- yaml: Outputs metrics in YAML format and saves it to a metrics.yaml file.
For detailed information on the metrics reported, please, visit the Nada Metrics section.
Example metrics usage
The following command runs the addition_test program and outputs detailed performance metrics in text format:
nada run addition_test --metrics text
- Metrics
- Nada program
- Test file
(.venv) ➜ nada-by-example git:(main) ✗ nada run addition_test --metrics text
Running program 'addition' with inputs from test case addition_test
Building ...
Running ...
Program ran!
Preprocessing elements:
Execution metrics:
Total duration: 1ms 160us 150ns
Total rounds: 0
Local protocols: (execution order)
Addition:
calls: 1
total duration: 114us 383ns (min: 292ns, max: 570us 208ns)
Online protocols: (execution order)
Outputs: {
"sum": SecretInteger(
NadaInt(
40,
),
),
}
loading...
loading...
Run and get metrics on the execution plan
To get detailed metrics on the program execution, including information on each protocol instance execution, we can use the flag --metrics-execution-plan
in addition to the command we see in the Run and get metrics section.
The result report includes the following information:
- The summary of metrics on the program execution.
- The detailed metrics on the protocol execution (instructions). This information is organized following the Execution Plan. If you want to know more about the execution plan, you can visit the Execution Plan section.
Example metrics on execution plan usage
The following command runs the addition_test program and outputs detailed performance metrics in text format:
nada run addition_test --metrics text --metrics-execution-plan
- Metrics
- Nada program
- Test file
Summary:
Preprocessing elements:
Execution metrics:
Execution duration:
431us 35ns
Compute duration:
total duration: 28us 894ns (min: 10us 701ns, max: 90us 242ns)
Total rounds: 0
Local protocols: (execution order)
Addition:
calls: 1
total duration: 16us 630ns (min: 6us 459ns, max: 49us 994ns)
Online protocols: (execution order)
Execution Plan:
Step #0:
Count of communication rounds: 0
Protocols:
addr(2) - Addition:
Execution step: 0
Total duration: 16us 630ns
Number of communication rounds: 0
Total message size: 0
loading...
loading...
Run and get protocols model text file
To execute a program and have a protocols model text file <test-name>.nada.protocols.txt
generated for your program, run with -protocols-text
or -p
nada run <test-name> -protocols-text
Example program and generated protocols model text file:
from nada_dsl import *
def nada_main():
party1 = Party(name="Party1")
party2 = Party(name="Party2")
party3 = Party(name="Party3")
a = SecretInteger(Input(name="A", party=party1))
b = SecretInteger(Input(name="B", party=party2))
result = a + b
return [Output(result, "my_output", party3)]
Here is the generated protocols model text file for the 3 Party addition program above:
Literals:
Inputs:
iaddr(0): A (sizeof: 1)
iaddr(1): B (sizeof: 1)
Outputs:
addr(2) (ShamirShareInteger): my_output
Execution plan:
Execution step 0 [Local]:
addr(2) - rty(ShamirShareInteger) = "ADD" iaddr(0) iaddr(1) # result = a + b -> main.py:11
Execution step 0 [Online]:
Run and get bytecode in text format
To execute a program and have a bytecode text file <test-name>.nada.bytecode.txt
generated for your program, run with -bytecode-text
or -b
nada run <test-name> -bytecode-text
Here is the generated bytecode text file for the 3 Party addition program above:
Header:
Party: Party1 id(0) # party1 = Party(name="Party1") -> main.py:4
Inputs:
iaddr(0) rty(SecretInteger) = Input(A) # a = SecretInteger(Input(name="A", party=party1)) -> main.py:7
Outputs:
Party: Party2 id(1) # party2 = Party(name="Party2") -> main.py:5
Inputs:
iaddr(1) rty(SecretInteger) = Input(B) # b = SecretInteger(Input(name="B", party=party2)) -> main.py:8
Outputs:
Party: Party3 id(2) # party3 = Party(name="Party3") -> main.py:6
Inputs:
Outputs:
oaddr(0) rty(SecretInteger) = Output(my_output) addr(2) # -> main.py:12
Literals:
Operations:
addr(0) rty(SecretInteger) = Load iaddr(0) # a = SecretInteger(Input(name="A", party=party1)) -> main.py:7
addr(1) rty(SecretInteger) = Load iaddr(1) # b = SecretInteger(Input(name="B", party=party2)) -> main.py:8
addr(2) rty(SecretInteger) = Addition addr(0) addr(1)
Test a program
To test
a program using nada
you need a test file, visit the Generate a test file section to see how to generate it.
Test in normal mode
To run all the tests in the project you can run
nada test
To run a single test you can run
nada test <test-name>
This will run the program associated with that test file and check if the output matches what's expected.
Test in debug mode
You can also run a test in debug mode, to know more about debug mode visit the Run in debug mode section. To run a test in debug mode you can run
nada test --debug <test-name>
Get a benchmark summary for a program
The Nada CLI benchmarking tool lets you run multiple programs and track their performance across multiple runs. It gives you a report that includes a summary of the metrics of the executions for each program. Visit the Nada Metrics section for more information about them.
This helps developers fine-tune their programs by showing where time is spent and how much data is transmitted during each protocol round.
Usage
Usage: nada benchmark [OPTIONS] [TESTS]...
Arguments:
[TESTS]... Names of the tests to use to benchmark the programs, if not specified benchmark all tests
Options:
--run-count <RUN_COUNT> How many times each test should be run [default: 1]
--message-size-calculation Measure protocol round size. Sizes are in bytes
-h, --help Print help
Example benchmarking
- Benchmark
- Nada program
- Test file
nada benchmark list_scan_linear_test
Running...
████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████ 1/1
╭────────┬─────────────────┬──────────────────┬────────┬────────────────────┬────────────────────────┬─────────────────┬──────────────────╮
│ Benchmark summary │
├────────┼─────────────────┼──────────────────┼────────┼────────────────────┼────────────────────────┼─────────────────┼──────────────────┤
│ Test │ Total Duration │ Compute Duration │ Rounds │ Total message size │ Preprocessing elements │ Local protocols │ Online protocols │
├────────┼─────────────────┼──────────────────┼────────┼────────────────────┼────────────────────────┼─────────────────┼──────────────────┤
│ test01 │ 4ms 453us 487ns │ 250us 291ns │ 46 │ 24kB │ 22 │ 2 │ 2 │
├────────┼─────────────────┼──────────────────┼────────┼────────────────────┼────────────────────────┼─────────────────┼──────────────────┤
│ 1 run(s) per protocol │
╰────────┴─────────────────┴──────────────────┴────────┴────────────────────┴────────────────────────┴─────── ──────────┴──────────────────╯
╭───────────────────────┬─────────┬───────────────────────╮
│ Preprocessing elements │
├───────────────────────┼─────────┼───────────────────────┤
│ Test │ compare │ equals integer secret │
├───────────────────────┼─────────┼───────────────────────┤
│ list_scan_linear_test │ 2 │ 20 │
╰───────────────────────┴─────────┴───────────────────────╯
╭───────────────────────┬───────────┬───────┬───────────────────────┬───────────┬───────╮
│ Local protocols │
├───────────────────────┼───────────┼───────┼───────────────────────┼───────────┼───────┤
│ Addition │ IfElsePublicBranches │
├───────────────────────┼───────────┼───────┼───────────────────────┼───────────┼───────┤
│ Test │ Duration │ Calls │ Test │ Duration │ Calls │
├───────────────────────┼───────────┼───────┼───────────────────────┼───────────┼───────┤
│ list_scan_linear_test │ 6us 624ns │ 20 │ list_scan_linear_test │ 5us 216ns │ 20 │
╰───────────────────────┴───────────┴───────┴───────────────────────┴───────────┴───────╯
╭───────────────────────┬─────────────┬───────┬────────┬────────────┬───────────────────────┬────────────┬───────┬────────┬────────────╮
│ Online protocols │
├───────────────────────┼─────────────┼───────┼────────┼────────────┼───────────────────────┼────────────┼───────┼────────┼────────────┤
│ EqualsSecret │ LessThanShares │
├───────────────────────┼─────────────┼───────┼────────┼────────────┼───────────────────────┼────────────┼───────┼────────┼────────────┤
│ Test │ Duration │ Calls │ Rounds │ Round data │ Test │ Duration │ Calls │ Rounds │ Round data │
├───────────────────────┼─────────────┼───────┼────────┼────────────┼───────────────────────┼────────────┼───────┼────────┼────────────┤
│ list_scan_linear_test │ 455us 890ns │ 20 │ 40 │ 19kB │ list_scan_linear_test │ 158us 16ns │ 2 │ 6 │ 5kB │
╰───────────────────────┴─────────────┴───────┴────────┴────────────┴───────────────────────┴────────────┴───────┴────────┴────────────╯
loading...
loading...