Filecoin Core Devs Meeting 34 Agenda

[kaitlin-beegle]

Filecoin Core Devs Meeting 34 Agenda

• Meeting Date & Time: Thursday, January 13 @ 16:00 UTC
• Meeting will be recorded.

Agenda

• Discussing updates, hurdles, questions, and challenges from Filecoin implementations:
  • Lotus
    • [TBD] Small protocol bug fixes?
  • Forest
  • Fuhon
  • Venus
• Updates from Filecoin Foundation
  • Confirming update timelines and FIP statuses (including on FIP0027)
• Q&A
  • v16 network planning
  • Direction of storage market

[jennijuju]

want to chat more about fvm , go/rust actors + ntwkv16 stuff - suggest folks to go over fvm blog post and repo before hands, and bring questions + concerns to the call . (~15 min)

[kaitlin-beegle]

^ditto to the above, especially want to discuss v16 scheduling. @raulk, do you plan to attend this meeting?

[anorth]

Suggested topic: Motivation and direction for storage market development (please contribute commentary back to the discussion online. I'm unable to attend the call)

[anorth]

I have published a first draft of the proposal for an explicit FIL+ subsidy, decoupling verified deal rewards from storage power.

https://github.com/filecoin-project/FIPs/pull/263

This is a complicated change, so I'm seeking some review and a sense of alignment to provide the mandate to follow through with the more detailed implementation and analysis. I believe it to be a critical step in enabling innovation in storage deals and markets with the FVM.

[kaitlin-beegle]

^Alex can't come because of the timezone. May want to discuss potentially hosting two Core Devs calls, the way FilecoinPlus structures its governance meetings.

[raulk]

Hey @kaitlin-beegle. I'll update async here for FVM+nv16, let me know if you really need me to join sync to discuss details/logistics. My plate is extremely full right now, but I’ll gladly join if the sync interaction is highly beneficial in this case.

There are two FIPs we expect to submit for nv16.

1. Filecoin Virtual Machine
2. Gas model changes in preparation for user-programmability

First FIP will likely come by the end of January. The second FIP depends on observations and data gathered from the canary experiment, and it’s a bit harder to predict.

In terms of implementation, FVM changes the approach since clients using the reference FVM need to do very little work.

• Implementation for the first FIP:

  • Lotus: by the time it is submitted, it will have been implemented in Lotus.
  • Forest: very easy to implement as no FFI is required.
  • Venus: can probably borrow the implementation patch from Lotus.
  • Fuhon: will need to create FFI shims, but the interop between C and Rust should be pretty straightforward.

• Implementation for the second FIP:

  • Entirely implemented inside the reference FVM, so no work from implementers.

I think the main consideration for nv16 will be how to conduct testing (i.e. testnets). I haven't done much thinking on this, and would like to discuss first before I bring something concrete up to Filecoin Core Devs, if that's OK.