Thinking about #htmx:

👍 all the #HTML gets generated on the server; the source of truth.

👎 every #DOM manipulation requires a network request.

👍 #declarative easier to reason about.

👎 app behavior declared in bespoke attribute DSL; harder to reason about.

👍 no need to duplicate #backend state on the #frontend.

👎 no offline support.

#webDev #web #webDevelopment #hypermedia #hateoas

@VamptVo

a) HTML attr based DSLs give me Angular flashbacks. It's declarative, but sometimes I think declarative is harder to grok than imperative.

b) if your UI change required a network request anyway (even if only for confirmation of success) then the performance is the same. but for minor UI state changes, you probably have a point.

@macdonst says that #HATEOAS can lead to network request bloat when you need to make a chain of multiple requests to get some resource.

Does #HATEOAS include #WebSockets? If so, how does one delivery #hypermedia over a #socket?

#html #htmx #http #webDev #web #javaScript #rest #api

How do you build an infinite scroller with #HATEOAS?

#REST #http #html #javaScript #api #htmx

@b0rk @randomgeek @hbuchel It’s funny how most “#RESTful” #WebServices have similarly missed the full point of @fielding’s actual #REST “just normal web things” style by not using hypertext as the engine of application state (#HATEOAS).

They’re just replaying all of #SOAP’s and #CORBA’s sins by relying on a fixed interface that is either revealed through documentation or an #InterfaceDescriptionLanguage like #OpenAPI.

@jorgealvarez @mxbck

I'm really confused as to why this is so novel to those who comment here. #JSON was not even a part of #REST: it is usually not *representing* anything and is *not* self-explanatory. Clients somehow have to catch up to API changes all the time, and backend can go chill??? This must be an obvious flaw with #JSONAPI.

And yet, the concept of #hypermedia responses, that #HTMX and #HATEOAS try to remind us of, seems alien to most. Oof, now that's the power of gaslighting in development circles! :rickwhoah:

i've been making a #rest #api lately.

real rest, you know, hypertext-as-the-engine-of-application-state #hateoas

it's kind of nice, actually. i like it.

did the half-whispered fever-dream of general (front-end) clients that can let a user navigate such an api ever lead anywhere?

I released the new version 2.0.2 of JSON:API for Spring HATEOAS, see https://github.com/toedter/spring-hateoas-jsonapi/.

This is a bug-fix release, based on Spring Boot 3.0.2.

#rest #api #jsonapi #spring #hateoas

The quest for #REST https://blog.frankel.ch/quest-for-rest/

#WebAPIs #HATEOAS #HAL

I released the new major version 2.0.1 of JSON:API for Spring HATEOAS, see https://github.com/toedter/spring-hateoas-jsonapi/.

This is a bug-fix release, based on Spring Boot 3.0.1.

#rest #api #jsonapi #spring #hateoas

@mogsie @mamund Cool! I have had a thought simmering in the back of my head for a few years now about using a PlantUML statechart to generate the stubs of a hypermedia-driven API. This sounds like much the same, only without the hypermedia “front-end”.

I believe the request/response dependency you describe can be decoupled with hypermedia by describing each response as a distinct type the client knows how to process, without knowledge of what it has processed before or what it is supposed to process next (iow: statelessness).

I try to describe this decoupling in the following article:

https://nordicapis.com/rest-state-machine-revisited/

I would love to know what you think.

#rest #hypermedia #hateoas

The 2.0.0-SNAPSHOT of "JSON:API for Spring HATEOAS" is now based on Spring Boot 3.0.0. Try it out before I release 2.0.0! See https://github.com/toedter/spring-hateoas-jsonapi

#rest #spring #hateoas #jsonapi

Since implementing my first API, it has taken me almost 15 years to actually write an article on Hypermedia and HATEOAS for a journal.

Maybe I should have done this earlier.

#REST #HATEOAS #Hypermedia