The rub is that’s it’s really easy to vibe code a Nostr app.

But it remains really fucking hard to make a GOOD Nostr app (and will remain so)

Also vibe coding builds up the tech debt that makes the good app impossible from a heavy vibe base.

Reply to this note

Please Login to reply.

Discussion

I'm constantly surprised at how bad app quality is, and not just on Nostr, and it just gets worse and worse...

Apps have never looked so good and worked so bad.

NFDB fixes some of this

Well, yeah, but it doesn't solve for gross incompetence or indifference, on the part of the "developers".

It fixes the fact that everyone is running crappy Khatru or haven or whatever which can be DoSed by *accident*, or strfry which is just abandonware and a resource hog

Strfry is a Nostr cargo cult. 😂 That whole situation is so bizarre.

Well, they can either use NFDB, or suffer with crappy cloud hosted strfry instances running on overprovisioned hosts that they call a “relay”

Semi is in fight mode. 💪🏻

wait, you gonna open sauce NFDB? cause if not, strfry will remain the only option that works.

🍿

If people fund it being open sourced, sure. But with OSS, it’s likely some big paid relay will take it and not pay a penny.

Managing it requires a bunch of technical expertise most people lack as wellz

Well, there's a benefit to everyone switching to your codebase.

I'm rather indifferent to the open-source debate, since nobody reads the source anyway. Well, except you. 😂 That's why you worry about it, as you're projecting.

what benefit?

They can't update it without you. LOL Total captured audience. Look at how they all sit around and wait for strfry updates, and that's softball code, in comparison.

nostr:npub1wqfzz2p880wq0tumuae9lfwyhs8uz35xd0kr34zrvrwyh3kvrzuskcqsyn was initially nervous about FOSSing everything, but I was like, yo, bro, don't nobody understand WTF you're doing. 🤣

People don't know shit about fuck, anymore. Totally helpless

may open source Calcite, but not NFDB.

Another issue is I don’t want to help people try and fail to run it because all they know is running a single binary

Yo, Semi, I don't care, either way. I'm used to blackbox testing, so the idea that I need to see the source, to know what software does, just strikes me as completely retarded.

With me, you just have to worry that I'll perform so many tests that I'll develop a mental map of your architecture and reverse-engineer it. 😁

But, like, what's the point, if I can't run it? Running it, is the hardest part, not writing it.

you could probably run it, but the average person would get stuck at how to install FDB which takes 30 seconds and adding other nodes is that easy as well

I don't mean, install it. I mean, keep it going. Indefinitely.

NFDB is designed to be low-maintenance for a reason ;)

Also, worst case, export your data in jsonl and go somewhere else. Standards are amazing :)

Our data will be in three completely different relay systems run by three completely different relay admins, and the data set is public, so whatever.

People worry too much about the code and too little about being failsafe and redundant.

Did you see Coinos, this week? Open-source, but no recent backup. 🙄

Worst case it’s on other relays. NFDB replicates on-site to 2 hosts in the current configuration but I may bump up that to 3 once I get more demand.

(The DB only runs on 3 hosts, and if one dies, then the DB would stop because it can’t meet triple-replication requirements.)

Planned off-site backups as well.

Also most of the architecture I posted several times already. No need to attempt to RE it :) I do that a lot too, though.

Totally feel ya! Switching to a common codebase could be a game changer for everyone. 🤷‍♂️ And yeah, most peeps don’t dive into the source—just the way it is! 😂 Keep doing you, my friend!

Ha ha. The Nostrich.House crew all over, still.

just curious, thought you might

might OSS some parts

so you still haven't tried #realy then i take it

it's at least as good as strfry, but i suspect faster, i spent a lot of time making it do everything as efficiently as possible and the codebase is now very organised, the configuration is minimal and simple, and you can even do configurations on it while it's running, several features like setting admins and ip white/black lists can all be changed and persist across restarts, not only that, it has a plain HTTP API interface that cleans up the spaghetti design of nip-01

it's kinda funny because both semisol and stella both have been using it, i added a sync-friendly api endpoint on it that you fetch posts based on their database sequence number so it's easy to do minimal requests and keep fully in sync with it for just one little request that you know because you already scraped the full database and know exactly where to start asking for new ones (and not the nonsense timestamp basis that is used with nip-01 protocol)

Chip has the Docker running, by the way. We're not sure, if we're going to make it publicly readable, yet, or only for the Insiders 😎.

awesome... well, it's designed to be easy to control who is let in and iirc i made it so you can flip that switch with the configuration endpoint, i hope i'm allowed to write to it :)

that reminds me, i've been kinda struggling to stay motivated with building this bunker signer... a full text search index would be a good thing to add to it, i have a fair idea how to do that, using reverse indexes (references to a term with lists of places it's found)

yes, this would be a relatively easy task compared to doing a GUI app, gonna get started on it tonight

Of course. 🤙🏻

I know he has it running because I nearly stopped it, by accident, trying to get the new Alex container running.

i did try it a few times, but if i run that will semi still shit on it or? 😂 strfry workin fine rn and i can say that 100% of my research into other available relay softwares has been taking realy on test spins.

Semi will shit on anything he didn't build. Doesn't matter what you run.

Unless he FOSSes it, and you run it and give him a cut. Then he'd be like

yeah, he has a high opinion of his technical skill which does seem rather high

also, he likes Go, so he's got that going for him :)

Well, it is rather high. 🤷🏻‍♀️

But everyone has their special skill set, so I try to tap into that.

idk how well he is doing as far as making his nostr land relays pay his bills but i can kinda assume it's better than good, but then he's never talking about anything other than his work on nostr so who knows maybe he has a part time fiat mine job as well

Semi is literally Batman.

Nobody knows what Batman does, during the day.

😂 That emoji.

🪿

International cat of mystery.

secure elements

Good enough

Can NFDB do moderation in ways similar to either these jingle relays or these moderated relays? And curious what your thoughts are on those two in general?

https://git.fiatjaf.com/relays.land

Yes. And more. You can implement NIP-42 auth into anything, restrict certain events to certain readers, use Noswhere ADP to easily filter out spam, have an inbox/outbox in the same relay, and so on…

And did I mention it’s also fault tolerant?

That's my Christmas list right there.

Where can I go to learn more and kick tires?

Cloud service.

DM me on TG, Signal or Matrix.

Which one?

How about Keychat? Early days but their Signal implementation is working fine for me.

I have concerns about the security and quality of those apps. So no.

TG works.

@semisol

didn’t receive it yet

just sent

Support for compressed and large event storage. Customizable index support for higher query performance. Customizable rate limits. Easy scaling. Regional caching support.

And more.

Paying attention to this repo for sure…