Replying to Avatar Garbage nsec

I think your argument is by far the strongest for what we have now (neutral labelling, flexibility, wide berth, better of a petri dish for new ideas...). The biggest issue for me is that this more ambiguous approach pushes too much of the cognitive load over to users. Many of these benefits accrue to the devs directly and to the users in a roundabout way. Better the opposite I think, have the benefits accrue to the users directly and to the devs in a roundabout way.

Also while I agree devs are motivated to work through the inconsistent UX, it becomes hard to channel that motivation to concrete action. You have 3 clients doing it one way, 3 clients doing it another way, and 3 clients yet another way. All 9 are hearing complaints from users and feeling motivated to work things out, but work things out how? Often it becomes a very tense and drawn-out game of paper scissors rock.

I think a lot of this falls into the class of problems for which no later decision can make up for the absence of an earlier decision. And lexicons do force those decisions much earlier.
Avatar
Niel Liesmons 9mo ago

Great write up sir!

I think we can get the best of both on Nostr. Withe the **onNostr** being the important part.

The central repo on GitHub = :110percent: the issue. As soon as you take that of the picture and imagine a Nostr spec on Nostr, there's opportunity to introduce the best of what lexicons bring.

I'm personally leaning towards:

1️⃣ Wikis as the best way to publish specs

2️⃣ App Releases as a potentially more suited place to reference which specs your app is built on

Reply to this note

Please Login to reply.

Discussion

Avatar
Niel Liesmons 9mo ago

That way I can:

- have my detailed Kind 9 Chat message "lexicon"

- document it in a wiki that doesn't need to be merged necessarily (yet still becomes more valuable, the more it is referenced/zapped/replied/accepted in communities/....)

- reference it transparently at the app-level without having to stipulate in each message what spec is being used (which comes with privacy issues etc...)

Avatar
Niel Liesmons 9mo ago

How would you do lexicons btw, in a way that's backwards compatible? nostr:npub1hyxredcavc6ruqgsf4wf4hmakpwnvefmzaspl7dja6a2sxlx0q3sxwtqnx

Avatar
Garbage nsec 9mo ago

Over there the protocol simply mandates both backwards and forwards compatibility. But often times the thing to is to reference another lexicon, or to create a sub schema.

So let's say in a lexicon called:

com.example.post

I have an author field, and for that field I want to reference the profile definition in another lexicon called com.example.user, I pop in:

"author": {"type": "ref", "ref": "com.example.user#/defs/profile"}

And that references:

"profile": {

"type": "object",

"properties": {

"handle": {"type": "string"},

"displayName": {"type": "string", "maxLength": 256},

"avatar": {"type": "string", "format": "uri", "nullable": true}

}

Then it's all just lego blocks and you can worry about them block by block.

But even if you have a breaking change and you come up with a new lexicon for it, all the other apps using the old lexicon will continue to use it, it can't be deleted, so you've not broken anything on their end.

Avatar
Garbage nsec 9mo ago

More here

https://www.pfrazee.com/blog/lexicon-guidance

Avatar
Niel Liesmons 9mo ago

Lo, they think they're messy :winkwithtongue:

Avatar
Garbage nsec 9mo ago

> 2️⃣ App Releases as a potentially more suited place to reference which specs your app is built on

Makes a lot of sense to me. Without that real-world context you can be forced into a kind of lowest-common-denominator-of-possibilities outlook.

Avatar
Niel Liesmons 9mo ago

True story!

Will pitch the idea to nostr:npub1wf4pufsucer5va8g9p0rj5dnhvfeh6d8w0g6eayaep5dhps6rsgs43dgh9 , but it's not priority now.

Step one = Getting the communities running where we can collab on these NIPs in the first place.

Avatar
Garbage nsec 9mo ago

For sure. If you get it up I'm happy to test sending join events with some throwaway nostr:npubs. This here web UI seems like it'd make testing pretty easy for a small test group.

https://nostrtool.com/