Core dev meeting #71

in #core9 days ago

Hello ! New month new core dev meeting.

Generated summary

This is more of a broad overview so you can see what topics were discussed but I heavily encourage you to listen to the thing

Technical Updates & Development Progress

Community Management & Notifications

  • Merged changes to prevent communities from having more than 100 moderators/owners/admins
  • Working on enabling error notifications that currently don't reach users (e.g., when setting too many moderators)
  • Will implement error notifications as a new notification type in users' notification lists
  • Plans to drop one notification table to save database space

External Partnerships

  • Waiting for feedback from Coinbase on mesh-api

API Stack Release Plans

  • New API stack version targeted for end of Q3 (September timeframe)
  • Currently in first phase testing, near feature-complete
  • Extensive collaboration between development teams being encouraged

HiveD Node Optimizations

  • Significant memory reduction: shared memory file reduced from 24GB to 6GB
  • Improved performance for both memory usage and replay speed
  • No longer requires RAM disk - can use NVMe drives directly for replay
  • Created script for optimizing virtual memory settings (requires root privileges)
  • Working on Swagger-based documentation for HiveD's API

Application Development

HiveMind Improvements

  • Enhanced muting functionality - now comprehensive
  • Added traditional full-text search using Gin indexes
  • Implementing semantic search (AI/vector-based search)
  • Plans to deploy to api.syncad.com for public testing

Balance Tracker (with McFarhat's team)

  • New features: balances, pending coin conversions, market orders
  • Added average transfer statistics and overall analytics
  • Monitoring recurrent transfers and trade history APIs
  • Working on rich list feature
  • Adding ability to track which account actually performed operations (vs. just affected accounts)

Block Explorer Updates

  • Sped up witness feed calculations (now real-time vs. every 10 minutes)
  • Improved permanent search functionality
  • Adding more data for display

Libraries & Tools

Wax Development

  • Python version now feature-complete and recommended for use
  • Improved transaction creation and broadcasting speed
  • API generator now supports both TypeScript and Python
  • Built-in health checker with multi-server support
  • Supports offline operations

Beekeeper

  • Essentially finished, releasing this quarter
  • Extensively tested through Denser and Clive applications

Web Components

  • Released embeddable web components for posts and comments
  • Collaborating with PeakD team on customized business websites
  • Working with Denser team on server-side rendering for SEO
  • Cross-team cooperation to avoid reinventing solutions

UI & Frontend Work

Block Explorer UI (McFarhat's team)

  • Implemented multilingual support for 12 languages
  • Added missing account details (last updated dates, etc.)
  • Improved witness voting history with pagination and search filtering
  • Health checker UI/UX improvements

Future Development & Collaboration

Team Cooperation Initiatives

  • Encouraging more interaction between siloed development teams
  • Meetings with PeakD team for cooperation opportunities
  • Suggesting Three Speak team collaboration
  • Successful ongoing collaboration with McFarhat's team on block explorer

Technology Considerations

  • Discussion about Java support for Android development
  • Current focus on completing Python Wax implementation
  • Future plans for Rust support
  • Worker Bee still in TypeScript, Python version planned

External Opportunities

Podcasting Integration

  • Ongoing efforts to get Apple to adopt Hive for podcast live notifications (PodPing)
  • PodPing provides cross-platform live podcast notifications via Hive
  • Would require minimal technical resources from Hive team
  • Could potentially integrate Hive into core Apple platform functionality
Sort:  

No longer requires RAM disk - can use NVMe drives directly for replay

What? I've never used RAM disk for replay nor sync, only using it when running benchmarks/stress tests.

That's what you get when you fully trust an AI summary :-)

The actual point was more that when tuned properly, you don't actually get much benefit from using a ram disk any more. You can get nearly the same speed from nvme + tuned vm settings (which prevents a lot of unnecessary writebacks to the disk during replay).