One of the many things we discussed at 34C3 was what we should focus on for the coming year. In particular, we wanted a roadmap that was clear about what we want to ensure we get done, vs what would be really nice to have, and be able to help onboard newcomers to either category. Here is what we came up with:
Priority: New crypto(graphy!)
Many of the current primitives and protocols still retain their original designs from circa 2005, and need improvement. We have had a number of open proposals for several years with ideas, but forward progress has been slow. We all agreed that this needs to be our top priority for 2018. The core components are:
- New transport protocols (to replace NTCP and SSU). See Prop111.
- New onion-encryption protocol for building and using tunnels.
- New NetDB datatypes to enable enhanced Destinations. See Prop123.
- Upgraded end-to-end protocol (replacing ElGamal).
Work on this priority falls into several areas:
- Writing proposals.
- Writing working implementations of them that we can test.
- Reviewing proposals.
We cannot release new protocol specifications across the entire network without work on all of these areas.
Nice-to-have: Code reuse
One of the benefits of starting the above work now, is that over the last few years there have been independent efforts to create simple protocols and protocol frameworks that achieve many of the aims we have for our own protocols, and have gained traction with the wider community. By leveraging this work, we get a "force multiplier" effect:
- We benefit from protocol designs, security proofs, and code written by others, reducing the amount of work we need to do for the same level of feature-completeness and security assurances.
- Work we do can be leveraged by other communities, increasing their interest in collaborating with us, and thinking about I2P as a whole.
Priority: Clearnet collaboration
On that topic, we've been slowly building interest over the last six months or so. Across PETS2017, 34C3, and RWC2018, I've had some very good discussions about ways in which we can improve collaboration with the wider community. This is really important to ensure we can garner as much review as possible for new protocols. The biggest blocker I've seen is the fact that the majority of I2P development collaboration currently happens inside I2P itself, which significantly increases the effort required to contribute.
The two priorities in this area are:
- Set up a project-run development forum that is accessible both inside and outside I2P.
- Set up mailing lists for review and discussion of proposals (possibly connected to the above forum).
Other goals which are classed as nice-to-have:
- Set up a usable git-to-mtn pathway, to enable us to effectively solicit clearnet contributions on GitHub while keeping the canonical dev environment on Monotone.
- Write a "position paper" that accurately explains I2P to academic audiences, and puts it in context with existing literature.
I expect that collaborations with people outside I2P will be done entirely on GitHub, for minimal friction.
Priority: Preparation for long-lived releases
I2P is now in Debian Sid (their unstable repo) which will stablilise in around a year and a half, and has also been pulled into the Ubuntu repository for inclusion in the next LTS release in April. We are going to start having I2P versions that end up hanging around for years, and we need to ensure we can handle their presence in the network.
The primary goal here is to roll out as many of the new protocols as we feasibly can in the next year, to hit the next Debian stable release. For those that require multi-year rollouts, we should incorporate the forward-compatability changes as early as we can.
Priority: Pluginization of current apps
The Debian model encourages having separate packages for separate components. We agreed that decoupling the currently-bundled Java applications from the core Java router would be beneficial for several reasons:
- It codifies the boundary between the applications and the router.
- It should make it easier to get these apps running with non-Java routers.
- It would enable third parties to create "I2P bundles" containing just the applications they want.
In combination with the earlier priorities, this moves the main I2P project more in the direction of e.g. the Linux kernel. We will spend more time focusing on the network itself, leaving third-party developers to focus on applications that use the network (something that is significantly easier to do after our work in the last few years on APIs and libraries).
Nice-to-have: App improvements
There are a bunch of app-level improvements that we want to work on, but do not currently have the developer time to do so, given our other priorities. This is an area we would love to see new contributors for! Once the above decoupling is complete, it will be significantly easier for someone to work on a specific application independently of the main Java router.
One such application we would love to have help with is I2P Android. We will be keeping it up-to-date with the core I2P releases, and fixing bugs as we can, but there is much that could be done to improve the underlying code as well as the usability.
Priority: Susimail and I2P-Bote stabilisation
Having said that, we do want to work specifically on Susimail and I2P-Bote fixes in the near term (some of which have landed in 0.9.33). They have had less work over the last few years than other I2P apps, and so we want to spend some time bringing their codebases up to par, and making them easier for new contributors to jump into!
Nice-to-have: Ticket triage
We have a large backlog of tickets in a number of I2P subsystems and apps. As part of the above stabilisation effort, we would love to clean up some of our older long-standing issues. More importantly, we want to ensure that our tickets are correctly organised, so that new contributors can find good tickets to work on.
Priority: User support
One aspect of the above we will be focusing on is keeping in touch with users who take the time to report issues. Thank you! The smaller we can make the feedback loop, the quicker we can resolve problems that new users face, and the more likely it is that they keep participating in the community.
We'd love your help!
That all looks very ambitious, and it is! But many of the items above overlap, and with careful planning we can make a serious dent in them.
If you are interested in helping with any of the goals above, come chat to us! You can find us on OFTC and Freenode (#i2p-dev), and Twitter (@GetI2P).