Dear nostr:npub160tqky4g2ezpcwnaxu60regxhfsch04gg709w0vq9k0dt7n0xn9qqe50js, please have sex with Iris and be as fast as it. Iris, you might want to leverage Jumble's awesomeness as well.
What’s better in Jumble?
Iris July 2025 Update. If you last used Iris in 2023, give it a retry! https://iris.to
New features:
- Feed editor! Edit feeds and create your own.
- Relay-based feeds
- Real-time updating feeds
- Show which relays an event came from
- Read-only mode (view as user / pubkey login)
- Improved popular feed, calculated locally (no more API calls)
- Performance Optimizations - Significant feed rendering and caching improvements
- Encrypted uploads in private chats
- Notification feed functional and visual fixes
- Configurable imgproxy
- Swipable image / video carousel
- Setting to disable specific reactions or the whole reaction bar for faster & calmer experience
- Lots of small fixes and UI improvements
- Binary serialization of social graph — 55% reduction in size. Distributed with a dataset of 30K users, 200K follows and 2K mutes in 1.4MB.
Double ratchet group chat & multi-device support WIP. Zaps not working great atm, fixing. Want to add nutzaps.
Tauri Android & desktop app available if you don't want to depend on iris.to, DNS, cloudflare, your ISP etc.
https://github.com/irislib/iris-client/releases
Iris subscriptions are now live. Get relay & blossom hosting and a shorter iris.to username: https://iris.to/subscribe
🚀 zapstore CLI 0.2.3 released
Various fixes, including one related to updating date on release.
Developers: I recommend updating: `zapstore update zapstore`
Or get from https://zapstore.dev/download if using standalone (no need to install it if just using `zapstore publish`)
Can you whitelist me on cdn.zapstore.dev?
zapstore publish tries to upload there and fails despite my zapstore.yaml:
blossom_servers:
Is there any info on the chunking standard?
There's not much info on it. Maybe it's using https://github.com/hzrd149/blossom/blob/chunked-blobs/buds/10.md ? BUD-10 looks workable, but there's no merkle tree definition yet.
We need some standard for chunking of videos and other large files.
That enables
1) verifying what you're getting before you've downloaded the entire large file,
2) downloading simultaneously from multiple servers or peers and
3) saved and scrollable livestreams — just publish the updated merkle root to your stream file every few seconds.
While we're at it, why not use blossom-compatible (sha-256) Bittorrent v2 merkle-trees https://blog.libtorrent.org/2020/09/bittorrent-v2/ — or maybe nostr:npub1h0rnetjp2qka44ayzyjcdh90gs3gzrtq4f94033heng6w34s0pzq2yfv0g soon?
I actually made this naive chunking for encrypted DM attachments larger than 4 MB. Not a merkle tree, but a single list of chunks a bit like torrent v1. Browsers don't have a streaming encryption API so had to chunk em.
We need some standard for chunking of videos and other large files.
That enables
1) verifying what you're getting before you've downloaded the entire large file,
2) downloading simultaneously from multiple servers or peers and
3) saved and scrollable livestreams — just publish the updated merkle root to your stream file every few seconds.
While we're at it, why not use blossom-compatible (sha-256) Bittorrent v2 merkle-trees https://blog.libtorrent.org/2020/09/bittorrent-v2/ — or maybe nostr:npub1h0rnetjp2qka44ayzyjcdh90gs3gzrtq4f94033heng6w34s0pzq2yfv0g soon?
Combining local LLMs and search indexes (such as Nostr databases) will be powerful. Like Grok, but sovereign and private. Collective conscious of humanity should not be controlled by CEOs and governments.
Consecutive calls to recalc while it's still in progress now return the same promise instead of starting a new operation
OpenAI o3 optimized the recalc method, should be faster now. Also changed queueMicrotask to setTimeout, which might block less.
Installed it, send me test messages.
Having kids is the best life extension technology.
Reticulum would work with BLE as well as LoRa, TCP/IP and other transports. Does chunking for files, real time audio etc.
I made recalculateFollowDistances async, it now processes in batches of 1000 and uses queueMicrotask
Recently watched Eyes Wide Shut, great movie.
Reticulum.network seems like meshtastic but more interesting, with potential to scale globally. Meshtastic https://www.youtube.com/watch?v=aBt56UpaQ0E