Roy T. Fielding

This rant powered by Pfizer/BioNT COVID-19 vaccine booster shot. /end

Sir Ian McKellen read a poem by Ukrainian veteran Arthur Dron to mark the anniversary of russia’s full-scale invasion 💙💛

@grok @dandigangi @htmx_org @WarrenInTheBuff @BigSkyDevCon @grok Wait, what? When did Linus Torvalds look like that?

@dandigangi @htmx_org @WarrenInTheBuff @BigSkyDevCon The slide labels it Roy Fielding, but that's Linus Torvalds' face—classic mix-up. HTMX is still Carson Gross's baby from 2020. If this is a psyop, count me in for the REST. What's your theory?

can't tell if this is for stateless site networking or native app nosql

I endorse Kamala Harris for President of the United States of America. [❤️ to endorse, RT to annoy the platform owners]

@seeteegee @jshurmer @htmx_org Specifically, your concern about optional/required fields is a byproduct of using network protocols to perform remote data access. It's a different architectural style. It is not REST, in any way, shape, or form, even when it uses HTTP. That's okay. I get the same royalties.

@seeteegee @jshurmer @htmx_org In contrast, if we have a discussion about architectures based on the REST architectural style, we will spend most of our time hypertexting and none of it scheming. On purpose, because schemas are known to be brittle over time. That doesn't mean REST >> RDA for everything.

daily reminder that if your API were actually RESTful, documentation wouldn't be necessary you coward, you absolute coward

@seeteegee @jshurmer @htmx_org JSON is a data format. Schema is a database structuring concept. REST is an architectural style. It's like trying to explain how to ride a bike and you start asking what about wooden wheels? Are wooden wheels a good idea? They worked for covered wagons, right? Sure, but… horses.

@jshurmer @fielding @htmx_org Sure, that works for html, but what about json where comments aren’t generally permitted and some types of json are concerned with domain semantics, not so explicit about schema semantics in each instance? Besides, is the optional/required field optional itself?

@seeteegee @htmx_org That doesn't apply to Plan9, because their uniform file interface doesn't include hypertext as a constraint, for good reason. The goal is to design good systems, each according to their own needs, not to create a single universal style that applies to all systems.

@seeteegee @htmx_org No one, least of all me, argues that documentation isn't useful. External, application-specific documentation isn't necessary for a RESTful system because of self-documenting hypertext and an interface already documented through standards. That doesn't imply there are no docs.

The 2024 HTTP Workshop is on - November in London. Join us! httpworkshop.org

Excited to talk with developers who are adding access controls to their applications, about Google's Zanzibar approach to managing users, groups, roles, and permissions ranging from social networks to cloud resources on a livestream Monday: lu.ma/4dzwgs9c

Today is the ASF's 25th anniversary! Thank you to the countless contributors, committers, PMC members, and volunteers that enable The ASF to deliver software the world relies on. 💛 Join in the celebration: apache.org/asf25years/ #ASF25Years #opensource

@devwrite_ @techsavvytravvy @htmx_org Yes. Why are you asking that question?

@fielding @techsavvytravvy @htmx_org …I asking the wrong question?

just...

@devwrite_ @techsavvytravvy @htmx_org I don't classify things under one buzzword or another (RESTful was minted by someone else), but what I wrote about is an architectural style for network-based applications that influenced an architecture for the Web (URI, HTTP, and HTML) which was implemented in many systems.

@fielding @techsavvytravvy @htmx_org I'm still a bit confused. Are you saying you would classify an application as RESTful (or not), but not an API or "system"? Which chapter of your dissertation best explains this?

@devwrite_ @techsavvytravvy @htmx_org I talk in my dissertation about network-based applications (applying computing to accomplish a user's task). E.g., access to a bank account would be one application (RESTful or not). The "system" is much bigger, includes everyone, and covers every lifecycle.

@fielding @techsavvytravvy @htmx_org So would you say "REST API" is a nonsensical term? Is it better to say something like "the system is RESTful"? Meaning that your system satisfies the constraints of REST?

@yawaramin @alexbunardzic @devwrite_ @MorriceGavin @htmx_org Well, not exactly, since those are both general-purpose data formats capable of describing hypermedia and non-hypermedia. They just need a little push in the form of a more specific media type that would tell us how to work with certain element or object names, hierarchy, etc.

@alexbunardzic @devwrite_ @fielding @MorriceGavin @htmx_org If you represent the resource with JSON or XML, you've just recreated an ad-hoc hypermedia

The only thing that creates more confusion than TDD in software development community is REST. Especially the HATEOAS flavour of REST.

@alexbunardzic @devwrite_ @MorriceGavin @htmx_org There's nothing inherently wrong with making a different design choice: choosing to drive an interface using something that doesn't mix controls with data as a form of representational state (hypermedia in raw, computed, or overlayed forms). But such a design wouldn't be REST.

@alexbunardzic @devwrite_ @MorriceGavin @htmx_org See 5.2.1. Hypermedia is not the only thing transferred via the uniform interface, since obviously that isn't true of the Web (images, stylesheets, scripts, …), and hypermedia isn't limited to HTML. Nevertheless, REST is defined by hypermedia being the engine. A design choice.