adding datagram and i2cp options to SessionOptions by sjlgasnier · Pull Request #11 · eepnet/yosemite

sjlgasnier · 2025-07-14T17:37:57Z

An humble proposal to resolve issue #10, adding as a bonus (most of the) i2cp options alongside the datagram ones, in case the former become needed in the future.

altonen

Thanks for the PR!

altonen · 2025-07-19T05:40:06Z

src/options.rs

+    /// IP address where the datagram socket should be bound to
+    /// 
+    /// Defaults to `0`.
+    pub datagram_host: u128,


This should be String and it should default to 127.0.0.1

In fact, datagram_host and datagram_port can be removed, they're automatically configured.

Aren't datagram_host and datagram_port needed for the two methods of #9 ? Unless they are also automatically configured at the datagram message level ?

datagram_host and datagram_port specify the address of the client's socket and they're sent in SESSION CREATE message to the router so it can relay datagrams received from the network to the client.

DatagramOptions should only have the following flags:

[FROM_PORT=nnn] # SAM 3.2 or higher only, default 0 [TO_PORT=nnn] # SAM 3.2 or higher only, default 0 [PROTOCOL=nnn] # SAM 3.2 or higher only, only for RAW sessions, default 18 [SEND_TAGS=nnn] # SAM 3.3 or higher only, number of session tags to send # Overrides crypto.tagsToSend I2CP session option # Default is router-dependent (40 for Java router) [TAG_THRESHOLD=nnn] # SAM 3.3 or higher only, low session tag threshold # Overrides crypto.lowTagThreshold I2CP session option # Default is router-dependent (30 for Java router) [EXPIRES=nnn] # SAM 3.3 or higher only, expiration from now in seconds # Overrides clientMessageTimeout I2CP session option (which is in ms) # Default is router-dependent (60 for Java router) [SEND_LEASESET={true,false}] # SAM 3.3 or higher only, whether to send our leaseset # Overrides shouldBundleReplyInfo I2CP session option # Default is true

Oops sorry I misread the docs indeed. Will delete.

altonen · 2025-07-19T05:40:20Z

src/options.rs

-    ///
-    /// Defaults to `2`.
-    pub num_outbound: usize,
+    pub inbound_qty: usize,


Suggested change

pub inbound_qty: usize,

pub inbound_quantity: usize,

altonen · 2025-07-19T05:40:33Z

src/options.rs

+    /// How many outbound tunnels does the tunnel pool of the session have.
+    ///
+    /// Defaults to `2`.
+    pub outbound_qty: usize,


Suggested change

pub outbound_qty: usize,

pub outbound_quantity: usize,

src/options.rs

altonen · 2025-07-19T05:41:26Z

src/options.rs

    /// Outbound-only sessions (clients) shouldn't be published whereas servers (accepting inbound
    /// connections) need to be published.
-    ///
+    /// 


Suggested change

///

///

altonen · 2025-07-19T05:42:15Z

src/options.rs

 }

+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct OtherI2CPOptions {


To remain consistent with the current code style, add line breaks between the options

src/proto/session.rs

altonen · 2025-07-19T05:49:57Z

src/proto/session.rs

-                match &self.options.destination {
-                    DestinationKind::Transient => {
-                        command += "DESTINATION=TRANSIENT ";
+                    "RAW" => {


It doesn't look like DESTINATION is specified for RAW. Could you please reduce code duplication between the three protocol kinds? It should match after parameters.options have been handled and only add FROM_PORT, TO_PORT, PROTOCOL and HEADER if the style indicates datagrams. Otherwise the different styles should be able to use the same code

…ions methods

sjlgasnier · 2025-07-19T17:34:33Z

I added two methods in datagram.rs that I believe will solve #9
EDIT: that will actually be implemented properly in a subsequent PR

I also deleted the fastreceive option after reading this exchange : eepnet/emissary#28

sjlgasnier · 2025-07-19T21:29:40Z

I believe the signature type datagram option should be added as well, as new post-quantum crypto types should eventually be approved, numbered from 12 to 20 as per proposal 169 : https://geti2p.net/spec/proposals/169-pq-crypto#signatures

altonen · 2025-07-20T07:45:06Z

src/asynchronous/session/style/datagram.rs

            .map_err(From::from)
    }

+    pub(crate) async fn _send_to_with_options(


The idea is to have a public function for Session<Repliable>/Session<Anonymous> like:

#[derive(Default)] struct DatagramOptions { from_port: u16, to_port: u16, } impl Session<Repliable> { pub async fn send_to_with_options( &mut self, buf: &[u8], destination: &str, options: DatagramOptions, ) -> crate::Result<()> { style::Repliable::send_to_with_options(&mut self.context, buf, destination, options).await } }

and the style::Repliable::send_to_with_options() would then fill FROM_PORT/TO_PORT from DatagramOptions. This allows overriding the source/destination port specified in SESSION CREATE message.

These functions must be available on both synchronous and asynchronous APIs

Ok I see, I will rollback these for now and will submit another PR afterwards.

altonen · 2025-07-20T07:46:13Z

src/asynchronous/session/style/datagram.rs

-            let socket = UdpSocket::bind(format!("127.0.0.1:{}", options.datagram_port)).await?;
+            let socket = UdpSocket::bind(format!("127.0.0.1:{}", 0u16)).await?;


My mistake, sorry. Please revert this change and keep datagram_port in SessionOptions. I forgot that the port for datagrams is taken from the config

altonen · 2025-07-20T07:47:18Z

src/options.rs

+    /// (ms) Idle time required
+    ///
+    /// Defaults to '1800000' ms (i.e. 30 minutes)
+    pub close_idle_time: usize,


Suggested change

/// (ms) Idle time required

///

/// Defaults to '1800000' ms (i.e. 30 minutes)

pub close_idle_time: usize,

/// Idle time required

///

/// Defaults to 30 minutes.

pub close_idle_time: Duration,

Is the comment correct? It's the same for both close_idle_time and reduce_idle_time

altonen · 2025-07-20T07:48:02Z

src/options.rs

+    /// (ms) Idle time required
+    ///
+    /// Defaults to '1200000' ms (i.e. 20 minutes)
+    pub reduce_idle_time: usize,


Suggested change

/// (ms) Idle time required

///

/// Defaults to '1200000' ms (i.e. 20 minutes)

pub reduce_idle_time: usize,

/// Idle time required

///

/// Defaults to 20 minutes

pub reduce_idle_time: Duration,

altonen · 2025-07-20T07:48:31Z

src/options.rs

+            crypto_low_tag_threshold: str::parse::<usize>("30").unwrap(),
+            crypto_ratchet_inbound_tags: str::parse::<usize>("160").unwrap(),
+            crypto_ratchet_outbound_tags: str::parse::<usize>("160").unwrap(),
+            crypto_tags_to_send: str::parse::<usize>("40").unwrap(),


Why are these str::parse()ed?

I was a bit tired aha. Fixed it.

altonen · 2025-07-20T07:48:54Z

src/options.rs

+            outbound_random_key: None,
+            should_bundle_reply_info: true,
+            close_on_idle: false,
+            close_idle_time: str::parse::<usize>("1800000").unwrap(),


Suggested change

close_idle_time: str::parse::<usize>("1800000").unwrap(),

close_idle_time: Duration::from_millis(1800000),

altonen · 2025-07-20T07:49:15Z

src/options.rs

+            lease_set_secret: None,
+            lease_set_signing_private_key: None,
+            reduce_on_idle: false,
+            reduce_idle_time: str::parse::<usize>("1200000").unwrap(),


Suggested change

reduce_idle_time: str::parse::<usize>("1200000").unwrap(),

reduce_idle_time: Duration::from_millis(1200000),

altonen · 2025-07-20T07:51:00Z

src/proto/session.rs

-                if !self.options.publish {
+                match parameters.style.as_str() {
+                    "PRIMARY" => {}
+


Please remove these line breaks between the styles

altonen · 2025-07-20T07:51:50Z

src/proto/session.rs

+
+                    "DATAGRAM" => {
+                        command += format!(
+                            "FROM_PORT={} TO_PORT={}",


Suggested change

"FROM_PORT={} TO_PORT={}",

"FROM_PORT={} TO_PORT={} ",

Whitespace needed at the end so the command is valid

altonen · 2025-07-20T07:52:08Z

src/proto/session.rs

+
+                    "RAW" => {
+                        command += format!(
+                            "FROM_PORT={} TO_PORT={} PROTOCOL={} HEADER={}",


sjlgasnier · 2025-07-26T05:56:05Z

I would open a PR in emissary to align with the renaming of num_inbound/outbound to inbound/outbound.quantity in yosemite. Very light refactoring.

altonen · 2025-07-26T08:05:32Z

I need to cut a new release for yosemite in order for emissary to use these changes and I would like to include both this PR and #9 for the next release. Feel free to make a PR if you want, it should be quite doable as Stream::connect_with_options can be used as an example.

sjlgasnier · 2025-07-26T08:14:55Z

Ah so you mean merging this PR will not automatically cause compatibility issues with emissary-cli>tunnel>client.rs for example ?
If so I will first tackle #9 then wait for the new yosemite release to adapt emissary (?)

altonen

Left some nits so that documentation looks consistent, otherwise LGTM

altonen · 2025-07-26T08:06:19Z

src/options.rs


+    /// Signature type.
+    ///
+    /// Default to '7' i.e. EdDSA_SHA512_Ed25519


Suggested change

/// Default to '7' i.e. EdDSA_SHA512_Ed25519

/// Default to `7`, i.e., EdDSA-SHA512-Ed25519

altonen · 2025-07-26T08:08:05Z

src/options.rs

+
+    /// If incoming zero hop tunnel is allowed
+    ///
+    /// Defaults to 'false'


Suggested change

/// Defaults to 'false'

/// Defaults to `false`.

altonen · 2025-07-26T08:08:34Z

src/options.rs

+    pub inbound_quantity: usize,

-    /// How many outbound tunnels does the tunnel pool of the session have.
+    /// Number of redundant fail-over for tunnels in


This is missing something? It's a half a sentence

This is how it is described in the I2CP doc. I will rewrite as /// Number of redundant fail-over for inbound tunnels.

True, sounds weird though. Maybe "Number of redundant, fail-over inbound tunnels" + same for the outbound case

altonen · 2025-07-26T08:09:41Z

src/options.rs

+    /// Number of IP bytes to match to determine if two routers should not be in the same tunnel. 0
+    /// to disable.
+    ///
+    /// Defaults to `0`.
+    pub inbound_ip_restriction: usize,


Suggested change

/// Number of IP bytes to match to determine if two routers should not be in the same tunnel. 0

/// to disable.

///

/// Defaults to `0`.

pub inbound_ip_restriction: usize,

/// Number of IP bytes to match to determine if two routers should not be in the same tunnel.

///

/// Defaults to `None`.

pub inbound_ip_restriction: Option<std::num::NonZeroUsize>,

altonen · 2025-07-26T08:10:09Z

src/options.rs

+
+    /// If outgoing zero hop tunnel is allowed
+    ///
+    ///  Defaults to 'false'


Suggested change

/// Defaults to 'false'

/// Defaults to `false`.

altonen · 2025-07-26T08:12:19Z

src/options.rs

+
+    /// Close I2P session when idle
+    ///
+    /// Defaults to 'false'


Suggested change

/// Defaults to 'false'

/// Defaults to `false`.

altonen · 2025-07-26T08:12:36Z

src/options.rs

+    /// (ms) Idle time required before closing session
+    ///
+    /// Defaults to '1800000' ms (i.e. 30 minutes)
+    pub close_idle_time: Duration,


Suggested change

/// (ms) Idle time required before closing session

///

/// Defaults to '1800000' ms (i.e. 30 minutes)

pub close_idle_time: Duration,

/// Idle time required before closing session

///

/// Defaults to 30 minutes.

pub close_idle_time: Duration,

altonen · 2025-07-26T08:13:11Z

src/options.rs

+
+    /// The encryption type to be used.
+    ///
+    /// Defaults to '4' i.e. ECIES-X25519


Suggested change

/// Defaults to '4' i.e. ECIES-X25519

/// Defaults to `4`, i.e., ECIES-X25519.

altonen · 2025-07-26T08:13:36Z

src/options.rs

+
+    /// Reduce tunnel quantity when idle
+    ///
+    /// Defaults to 'false'


Suggested change

/// Defaults to 'false'

/// Defaults to `false`.

altonen · 2025-07-26T08:13:54Z

src/options.rs

+    /// (ms) Idle time required before reducing tunnel quantity
+    ///
+    /// Defaults to '1200000' ms (i.e. 20 minutes)
+    pub reduce_idle_time: Duration,


Suggested change

/// (ms) Idle time required before reducing tunnel quantity

///

/// Defaults to '1200000' ms (i.e. 20 minutes)

pub reduce_idle_time: Duration,

/// Idle time required before reducing tunnel quantity

///

/// Defaults to 20 minutes.

pub reduce_idle_time: Duration,

sjlgasnier · 2025-07-26T09:48:22Z

Fixed the nits, noticed I had missed a lease set option, and then tried to reorder the options in a somewhat logical order.

sjlgasnier · 2025-07-26T09:51:41Z

Should I define outbound/inbound_len as std::num::NonZeroUsize for optimization (emissary does not support 0-hop tunnels) ?

altonen · 2025-07-26T10:21:10Z

Should I define outbound/inbound_len as std::num::NonZeroUsize for optimization (emissary does not support 0-hop tunnels) ?

yosemite can be used with java/i2pd which support zero-hop tunnels and emissary should eventually support them as well so no need

sjlgasnier added 3 commits July 14, 2025 19:21

adding datagram and i2cp options to SessionOptions

04b0515

typo session.rs

5078ae2

closer to the original code

7e1d951

altonen reviewed Jul 19, 2025

View reviewed changes

adding datagram and i2cp options to SessionOptions with send_with_opt…

e18da66

…ions methods

altonen reviewed Jul 20, 2025

View reviewed changes

sjlgasnier added 3 commits July 21, 2025 08:34

adding datagram and i2cp options to SessionOptions

ff70007

adding datagram and i2cp options to SessionOptions

502267d

nit

5d39c4c

altonen approved these changes Jul 26, 2025

View reviewed changes

adding datagram and i2cp options to SessionOptions

ffad8dc

altonen approved these changes Jul 26, 2025

View reviewed changes

altonen merged commit 77357d2 into eepnet:master Jul 26, 2025
1 check passed

altonen mentioned this pull request Oct 19, 2025

Add datagram options to SessionOptions #10

Closed

		let socket = UdpSocket::bind(format!("127.0.0.1:{}", options.datagram_port)).await?;
		let socket = UdpSocket::bind(format!("127.0.0.1:{}", 0u16)).await?;

	close_idle_time: str::parse::<usize>("1800000").unwrap(),
	close_idle_time: Duration::from_millis(1800000),

	reduce_idle_time: str::parse::<usize>("1200000").unwrap(),
	reduce_idle_time: Duration::from_millis(1200000),

	/// Default to '7' i.e. EdDSA_SHA512_Ed25519
	/// Default to `7`, i.e., EdDSA-SHA512-Ed25519

	/// Defaults to '4' i.e. ECIES-X25519
	/// Defaults to `4`, i.e., ECIES-X25519.

Conversation

sjlgasnier commented Jul 14, 2025

Uh oh!

altonen left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sjlgasnier commented Jul 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sjlgasnier commented Jul 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

altonen Jul 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sjlgasnier commented Jul 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

altonen commented Jul 26, 2025

Uh oh!

sjlgasnier commented Jul 26, 2025

Uh oh!

altonen left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sjlgasnier Jul 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

sjlgasnier commented Jul 19, 2025 •

edited

Loading

sjlgasnier commented Jul 19, 2025 •

edited

Loading

altonen Jul 20, 2025 •

edited

Loading

sjlgasnier commented Jul 26, 2025 •

edited

Loading

sjlgasnier Jul 26, 2025 •

edited

Loading

altonen Jul 26, 2025 •

edited

Loading

sjlgasnier commented Jul 26, 2025 •

edited

Loading