joystream/reports/archive/2.md

Runtime Spec Proposal Discussion

Author

Bedeho

Participants

• Jens
• Alex
• Dagur
• Matthew
• Martin
• Bedeho

Time and place

3rd May 2019, 11:30am GMT+1

Topics

Bedeho presents and approach for how to do runtime specification up front with the goal of making internal coordination clearer, more error free and ultimately faster by reducing amount ambiguity and mistakes.

Proposal

Members Module

Conclusions

Bedeho post-mortem opinion

No one was terribly excited at the prospect of doing this sort of specification up front. The general feeling was that it was too detailed, and would take so much time to arrive upon up front. Some wanted a higher level description instead, while other suggested simply diving straight into implementation, but instead doing it in stages where the skeleton of a runtime is reviewed by someone else before full implementations are made. While I can appreciate the concern, I am still worried that we will end up shipping less secure/error free code and that our internal shared understanding of what is being built is going to be unreliable - slowing us down. None the less, I am happy to try this higher level approach on the runtime for now, as I am effectively a lone voice on this - and there are so many unknowns. I am not sure where this preference for more informality leaves us on for other nodes/protocols, but we can revisit this. This needs to be discussed further with Mokhtar, who had to leave early on the call.

Note

If any participant wishes to add their own post-mortem, please open a PR.