# CID Congress #5 - Agenda and Notes

20 Nov 2025
Event link: https://luma.com/5n563q8d
Link to last sessions's notes:

Recording (available for 2 weeks):

## Participants / githubs / socials :

Jon Geater
Benjamin Barber
Alex
Basile
Daniel Mantei
lidel
Steve Vandevelde
Sumanjeet Kumar
Wes Biggs
Volker
Mosh
Robin

## Reference links (to check out before the call):
-

## Agenda:

- Administratrivia
- This is every two months on the 3rd Thursday (next one is Jan 15th). We now have a calendar for this series: https://luma.com/cid-congress.
- Send us topics & demos!
- You might also enjoy DASLing Working Group (monthly meeting), ping [e-mail] for invite.
- DASL Updates (Robin Berjon & Juan Caballero)
- Spec changes
- Issues of interest
- Data Prep Tooling (Jon Geater)
- ATGeo Update (Volker Mische)
- CID Profiles (Mosh Lee)
- What do you want to see in 2026?

## Notes

- DASL Updates
- https://datatracker.ietf.org/doc/draft-caballero-cbor-cborc42/ was submitted as DRAFT to IETF
- Waiting for CBOR WG to solidify the format for CBOR profiles to be documented formally, and then we'll update the above draft
- RASL - changed the scheme and switched to query string hints
- MASL - made naming more coherent around content types and media types
- CIDs are now fixed length, to make it easier to write parsers and
- New suggestions for syntax to use in JSON, not required
- New spec coming soon on verifiable ranges & range requests
- A few open issues, check them out if you're interested and/or have an opinion: bignum support

- Jon Geater (CTO, FIDL)
- A lot of what I do is Filecoin-oriented, but driven by interoperability and efficiency for the full IPFS ecosystem
- Commons Data: information that has a collective benefit, that we want to ensure can survive for the long term
- Needs interop to simplify life for onboarders and users (writers & readers). JG not sure he wants to call it a standard but it helps with interop.
- TOADS v1 - includes manifest spec, tooling, and example directories
- Manifest has 2 parts: super manifest + sub-manifests
- Models a lump of files as a coherent single dataset, e.g. set of experimental results.
- Autogenerate metadata as much as possible, but not everything
- Very centred on CAR because Filecoin does that, but could be different.
- Roadmap:
- Onchain-ness
- Non-public data
- Different prep - non-CAR v2, integrating things like MASL/DASL
- Non-TOADS - what else is needed to make data discoverable for other use cases?

#### Discussion:

Benjamin Barber: LAION-5B (https://laion.ai/blog/laion-5b/) has 5 billion images, will that break your use case because of the number of items in the index?
JG: Theoretically no, but in practice probably. We've tested to a million file nodes, 3-4 PB datasets.
BB will connect with JG by email

Robin: Why CARv2?
JG: Not an opinionated choice

Wes Driggs (Project Liberty): Isn’t the goal of DRISL to remove the need to have separate cid and hash fields? (Obviously co-evolving with your work here)

Robin: Not sure MASL is a good fit here, you can use it but doesn't add a lot for datasets.
JG: What about a project for software toolchain resilience?

- ATGeo WG Update (Volker)
Slides: https://hackmd.io/@vmx/cid-congress-atgeo-wg

- Defining a lexicon for places
- External off-protocol gazeteer services
- Safety/privacy considerations
- Collecting use cases
- #wg-atgeo tag on the lexicon.community discourse: https://discourse.lexicon.community/tag/wg-atgeo
- #lcwg-atgeo-gazeteer channel on ATProto Touchers Discord

Manu: As an app developer, how can we use a lexicon in a project? Is there an API?
Volker: Still defining the lexicon.

- CID Profiles (mosh)
- https://github.com/ipfs/specs/pull/499
- Defines all the parameter settings that yield a given CID

Steven: this is my use case for the config thing you mentioned earlier in case you're interested: https://tangled.org/tokono.ma/diffuse-artifacts/

Basically, allows the user of a web app to pick a version. Based on CIDs.

- Misc
- Basile shared https://www.hytradboi.com/ - cool conference and cool format

- Tiktok is looking for someone to explain to her & LF folks to explain libp2p & IPFS, in context for mobile robotics - p2p instead of hitting. Radio access networks, towers might do some caching. Smart geo routing to get data closest to the people. Manu? Available now after Devconnect events.

- py-libp2p is close to feature-complete
- BB - using multiple tools that use multiformats, but all different versions. Python: had to import 3 different versions of multiformats and then namespacing them.
- Manu: that has been resolved. We now have a pypi version for mf, and you can invoke from there.

Wishlist:
- Daniel Mantei - more Rust (&WASM)
Coming from server development, a lot of these libraries are a pain to bring into Rust projects. Can be done via FFI, but it's still tortuous

https://github.com/n0-computer/dasl
https://github.com/dariusc93/rust-ipfs

- BB - wants a babysitter service

-