semisol

👨‍💻

Public Key

npub12262qa4uhw7u8gdwlgmntqtv7aye8vdcmvszkqwgs0zchel6mz7s6cgrkj

NIP-05 Address

semisol@nostrpurple.com

NIP-57 Address

zap@semisol.dev

Profile Code

nprofile1qqs99d9qw67th0wr5xh05de4s9k0wjvnkxudkgptq8yg83vtulad30gpp4mhxue69uhkummn9ekx7mqpz3mhxue69uhhyetvv9ujuerpd46hxtnfdudt29ct

Publishing to

nos.lol

relay.damus.io

nostr-2.21crypto.ch

Author Public Key

npub12262qa4uhw7u8gdwlgmntqtv7aye8vdcmvszkqwgs0zchel6mz7s6cgrkj

Show more details

Last Notes

2025-06-14 23:17:08

- reply

nostr.land

2025-06-14 23:12:13

The porn spam is being coordinated by the same people that ran the previous reply spam bots as they are followed by several accounts of the previous spammers.

2025-06-14 23:05:28

- reply

1. The media has been removed (I believe)
2. If any client supported the NB moderation headers, they would obviously see it is flagged as 99.7% NSFW: "X-Moderation: SAFE:0.000,QUESTIONABLE:0.003,UNSAFE:0.997"
3. working on a thing to scan all media on Nostr (h/t @npub137c…k37w): https://i.nostr.build/vaRSZTHPs38sxTdQ.png

2025-06-14 22:01:53

- reply

Receive

2025-06-14 21:51:59

- reply

you have skipped the in game currency black market directly to real money, congrats

2025-06-14 21:50:28

- reply

It does in fact work

2025-06-14 20:06:22

- reply

they are not sure if it is real

2025-06-14 20:02:56

- reply

For best results, disable conflict checking

2025-06-14 15:51:48

- reply

DMs

2025-06-14 14:57:53

- reply

I prefer vibe coding software that doesn’t even exist.
(code sold separately)

2025-06-14 14:38:39

- reply

I do not use AI generated code because it doesn’t have a place in critical Nostr infrastructure.

2025-06-14 14:34:21

- reply

Also, did I tell you that yesterday, I unplugged a server forcefully for about half an hour and the relay didn’t go down? 🤭

2025-06-14 14:32:55

- reply

Transparency matters.
The Event Cannon pipeline relied on CF which helped with ratelimiting.
To prevent this again, I will be setting up a queue for handling broadcasts and have fallbacks.

2025-06-13 20:51:18

- reply

The answer is obviously nostr.land. :)
While @npub1hu4…c46u has their own software that’s it. It runs on MongoDB which is known to be very slow, and they have no value proposition except paid relay.
Wine is a set of duct taped relays running on strfry, and you’d need 4 relay connections to only somewhat replicate the functionality of Nostr.land.
And Nostr.land offers convenient import/export and runs NFDB.

2025-06-13 17:23:59

- reply

It goes through a custom event processing pipeline and within 1-2 seconds you’ll see it on aggr.nostr.land.
Seconds, not minutes.
With NFDB the query performance is significantly better as well, and allows faster queries on large datasets.

2025-06-13 16:57:46

- reply

Currently trying to fully get the spam filtering down. The benefit over filter.nostr.wine is that it reads from hundreds of free and paid relays, compared to the less than 10 free relays they read from.

2025-06-13 16:56:46

- reply

Yes, I mentioned that last week.
This is only a temporary limitation because this is a beta service.

2025-06-13 06:44:04

About 229 event broadcasts failed yesterday during the outages. Writing to Nostr.land was otherwise unaffected
https://status.noswhere.com/issues/2025/06/ec-failures/

2025-06-13 02:11:09

I have seen the horrors that is the Cloudflare product ecosystem.
At any given moment in time, 50% of products run in degraded mode, and any 5% are near-down or unusable.
Not much different than Nostr ship turds fast mindset. Any day now they will have a catastrophic incident
CF and GCP, avoid unless your life depends on it
#note1daf…4q3l

2025-06-13 01:59:27

https://i.nostr.build/NP06UzZ4NW8QL8h5.jpg

2025-06-13 01:49:04

another day, more cloudflare services showing they are in fact polished turds

2025-06-12 21:01:23

- reply

But what was the leverage

2025-06-12 14:21:03

- reply

prism

2025-06-12 13:47:53

and cashu is the worst for this
#note1gwl…ueuq

2025-06-12 13:43:37

- reply

a changelog is needed

2025-06-12 13:14:20

- reply

ser blastr has happened

2025-06-12 07:20:37

- reply

No.
But the hosted pricing will be available soon.

2025-06-12 06:33:28

- reply

Nostr.land will soon only support LLM queries and all exchanges will be fed through LLMs before being converted to unoptimized SQL

2025-06-11 18:09:33

- reply

The same thing will happen to many relays

2025-06-11 16:52:29

- reply

What I meant was write your own Git remote fully instead of somewhat relying on the git upload-pack and similar utils

2025-06-11 16:47:38

- reply

If you ever want to do that, Git is not much different than a key-blob store

2025-06-11 16:27:58

- reply

Ah ok.
For NWA you’d need to use the provided KV for storage. And for blobs, put that in an object storage provider

2025-06-11 16:19:14

- reply

Local storage for blobs or something like S3/Blossom?

2025-06-11 16:10:52

- reply

What does it do exactly?

2025-06-11 16:07:15

- reply

Is that the thing you mentioned on NWA-1? Didn’t understand the context.

2025-06-11 15:51:50

- reply

it is somewhat tedious to set up, but it pays off a lot, it is basically impossible to break the DB

2025-06-11 15:33:57

- reply

1. download the suitable release from https://github.com/apple/foundationdb/releases/tag/7.3.63 (you need both -clients and -server for servers)
2. install with your package manager
3. edit /etc/foundationdb/foundationdb.conf
- set public-address = <ip>:$ID
- set machine-id to hostname
4. systemctl start foundationdb
ensure the FDB ports are firewalled
edit /etc/foundationdb/fdb.cluster on the first node, replace 127.0.0.1 with node IP
fire up fdbcli and run `configure ssd-redwood-1`
repeat on all nodes the installation, edit foundationdb.conf, BUT copy /etc/foundationdb/fdb.cluster from the main node
after you added all nodes run `coordinators auto`
if you want access control with mTLS instead that requires some additional setup

2025-06-11 15:27:30

- reply

when running FoundationDB?

2025-06-11 11:06:46

- reply

No, I am busy with working on relays 🤣 but nice to see it’s included.
Just don’t push it to release yet

2025-06-11 11:00:55

- reply

relays

2025-06-11 00:47:40

- reply

Preimage is deprecated ser

2025-06-10 23:35:07

- reply

Analyzing the power consumption of chips can allow leaking the full internal state of the CPU

2025-06-10 23:31:47

- reply

Technically you could use the EMI from the high-speed communication lines as an output

2025-06-10 23:28:15

- reply

The network is the output

2025-06-10 23:15:14

- reply

@npub1ye5…knpr fix this, it works on Nostrudel

2025-06-10 23:14:25

- reply

No one checks it. I don’t think many even check the amount tag, or even the zap service pubkey

2025-06-10 23:13:34

- reply

The ownership of the Rust user is returned to the hill queue by the turbine which is concurrency optimized. The trapdoor only passes ownership to the turbine and does nothing else

2025-06-10 23:12:17

- reply

posting this while rewriting parts of my codebase in Rust 🤣

2025-06-10 23:11:05

- reply

interactions after posts matter more than the posts themselves

2025-06-10 22:43:18

- reply

GH did not support detaching forks from the main repo for ages
New feature apparently

2025-06-10 22:41:29

- reply

and a monitor

2025-06-10 22:37:19

https://i.nostr.build/khyVu7Q901BPfGwV.png

2025-06-10 21:49:29

https://i.nostr.build/jQSsyOFq321xV6xs.jpg

2025-06-10 15:54:29

- reply

The same reason the Lightning port is 9735

2025-06-10 15:52:52

- reply

128169

2025-06-10 14:38:38

- reply

If you can read an NFC tag you can read a modern smart card

2025-06-10 14:20:47

- reply

For $1 more you can get about 50x more security

2025-06-10 14:05:06

- reply

Okay so it’s basically an overly complicated way of saying cloneable magstripe

2025-06-10 13:38:08

- reply

Or to phrase it better
Is it possible to read the tag without authentication and does having the data on the tag provide full access to the safebox

2025-06-10 13:17:58

- reply

then they can self host the backend

2025-06-10 07:25:18

Any sufficiently large distributed system is indistinguishable from a Matryoshka doll

2025-06-10 01:04:01

- reply

There’s also the fact that CFW is a JS runtime and would not be able to deliver performance anywhere close to solutions based off of even Java.
And lastly, D2 is based off of SQLite.
This puts a hard throughput limitation for the same cost, as SQLite optimizes for simplicity and portability, not performance.

2025-06-10 01:00:41

- reply

That’s a bold claim.
First of all, for the same load, it would cost significantly more to handle it with Nosflare than cloud compute.
Second, Nosflare would be crushed completely by strfry or NFDB in terms of write throughput, or the total amount of data that could be handled.

2025-06-09 13:47:37

- reply

this is from almost 2 years ago

2025-06-09 01:45:18

- reply

With Phoenix/Zeus you can enjoy force closes and channel fees, transferring every sat you earn from zaps to miners and/or LSPs, before you could even spend 1.

2025-06-09 01:42:31

low quality and vibecoded bitcoin apps need to go in the shredder before they cause everyone to lose all their money and ruin bitcoin's reputation

2025-06-09 01:05:54

- reply

They are if they want to offer a service to end users.

2025-06-09 01:03:43

- reply

can’t wait to get all my zaps consumed by force close and channel open fees

2025-06-09 00:49:31

- reply

1. Being FOSS does not release the dev of obligation to make secure products
2. If you make garbage and you reasonably could have known it was garbage, do not get upset if it gets called garbage and the dev an idiot

2025-06-09 00:45:06

- reply

update I tried this and now I have become a shift register

2025-06-09 00:04:18

Coinos has violated many basic operational practices at this point.
They did not have real-time replication of critical transaction data.
They stored user nsecs without a KMS system like OpenBao.
They did not have countermeasures against account breaches being able to extract confidential info.
Get your sats the fuck out of Coinos and don’t ever come back.
#note1nqs…2gs2

2025-06-08 23:29:32

- reply

why is my cluster not working ser

2025-06-08 23:26:38

- reply

kubectl delete node master-1 --force

2025-06-08 20:10:58

- reply

The solution is that the Blossom software should be doing CORS by default

2025-06-08 16:01:06

- reply

This does not require any physical contact and I thought we were trying to be better than fiat

2025-06-08 15:56:31

- reply

And then as fast as it is complete it becomes an unmaintainable mess

2025-06-07 20:57:59

- reply

Dumb Nostr clients

2025-06-07 20:48:54

- reply

Because productivity matters

2025-06-07 19:36:49

shit I deleted sources.list

2025-06-07 16:57:57

- reply

any grant over 20M sats to any Nostr dev has usually ended in disaster

2025-06-07 16:19:50

"No changes. Your infrastructure matches the configuration." 🥳

2025-06-07 16:00:54

- reply

Good because I would honestly not be surprised of such shit from OSMU

2025-06-07 16:00:08

- reply

What

2025-06-07 15:59:27

- reply

The one thing that might piss you off is if err != nil {} but I find that in most cases, either I can replace it with a panic-if-error or it has to have special handling anyway.

2025-06-07 15:58:14

- reply

My description isn’t really accurate either.
The build systems is sane unlike whatever the fuck Java and C++ are.
Also native monolithic binaries are a plus for everything except size.

2025-06-07 15:55:33

- reply

I am currently looking into Rust for some DB code that does high-performance processing.
Trying to optimize the GC would be about as much effort as just using Rust.
About C, it may as well be human readable portable assembly. Great for embedded platforms that I’m working with, otherwise something I don’t want to touch unless necessary

2025-06-07 15:50:40

- reply

Go fills the gap between C++ and Java.

2025-06-07 15:34:34

- reply

Yes.
You can get your account subscriptions merged too. There will probably be some sort of limit to keep it reasonable
DM me your npubs and I’ll figure it out
(if you have both a big personal + business account, you have to pay separately, but if it is small things then it’s fine)

2025-06-07 15:27:44

Multi-account support for nostr.land will be added soon 👀
#note1vnu…hwjq

2025-06-07 15:13:36

gm Nostr
things will be shipped

2025-06-07 14:49:38

- reply

It is an application optimized transaction database

2025-06-07 13:02:35

- reply

It works with Nostr apps instead of being a proprietary interface

2025-06-07 13:02:05

- reply

As a general relay

2025-06-06 23:50:52

- reply

holy fuck the HDR

2025-06-06 23:02:05

- reply

did not test

2025-06-06 22:44:21

- reply

I somehow managed to kill the etcd instance and cause the cluster to permanently fail because I restarted masters too many times in a test deployment.
I am honestly reconsidering my plan to use etcd for a few services

2025-06-06 22:42:34

- reply

k3s does have etcd for HA, if you want to suffer

2025-06-06 21:06:24

- reply

Yes

2025-06-06 21:02:06

- reply

About the point with Proxmox, that was on metal *but* cattle-type workloads.
Also yes it took me a few hours to spin up a K3s cluster (primarily reading the docs) and only an hour to then deploy a multi-node Pulsar and ClickHouse cluster

2025-06-06 20:55:54

I may have drank too much of the cloud juice
- Proxmox is not ideal for large-scale deployments with cattle
- K8s is complex but the complexity pays off 100x
- Route 53 just works
- CF Workers is good, sometimes
- R2 and B2 are the best object storage
- Hetzner is great for a lot of stuff
- EC2 as usual is a scam