the idea of a thoughts page was originated by maren, who made a script for generating thoughts pages. thoughts.page is a way of lowering the barrier of entry to putting thoughts on the internet for people who don't want to or don't know how to set up a script to do it.
Good use case for the application of the principles in A New Publishing Discipline.
Primary program modules
This is sort of a failing of the code-as-content thing that we're going for here. Take a page from GPE.
data can be exported from airtable, but the logic and UI can't
Unlike conventional blogs which are read-only this blog contains active code, so you can click on any of the links in this tiddler and see the code and see how the system behaves.
the process for putting something on the internet to just be a git push and trust that the machine will just take care of it
Are you limited to PHP?
No, but further: the question (about being "limited") presupposes something that isn't true.
If you're doing PHP here, you're doing it wrong—unless the PHP application is written with great care (i.e. unidiomatically) and has some way to reveal its own program text (as first-class content). Otherwise, that's a complete failure to avoid the "elsewhere"-ness that we're trying to eradicate.
who hosts that?
Answer: it's hosted under the same auspices as the main content. The "editor" is first-class content (in the vein of ANPD); it's really just another document describing detailed procedures for how the site gets updated.
the former allows me to give an URL to a piece of code
But you're not! When you wield PHP like this, there is no URL for the piece of code per se—only its (potentially fleeting) output—unless you take special care to make that piece of code available as content otherwise. PHP snippets are just as deserving of a minted identifier issued for them as, say, JS and CSS resources are—perhaps even just as deserving as the content being served up on the site, but PHP actually discourages this.
My argument for the use of the Web as a medium for publishing the procedures by which the documents from a given authority are themselves published shares something in common with the argument for exploiting Lisp's homoiconicity to represent a program as a data structure that is expressed like any other list.
There are traces here as well from the influence of the von Neumann computational model, where programs and data are not "typed" such that they belong to different "classes" of storage—they are one and the same.
Many of the items in the docuverse are not static, run-of-the-mill materials, i.e. unformatted text, graphics, database files, or whatever. They are, in fact, executable programs, materials that from a docuverse perspective can be viewed as Executable Documents (EDs). Such programs run the gamut from the simplest COBOL or C program to massive expert systems and FORTRAN programs. Since the docuverse address scheme allows us to link documents at will, we can link together compiled code, source code, and descriptive material in hypertext fashion. Now, if, in addition, we can prepare and link to an executable document an Input-Output Document (IOD), a document specifying a program's input and output requirements and behavior, and an RWI describing the IOD, we can entertain the notion of integrating data and programs that were not originally designed to work together.
GitHub is a foreign CMS
The final keystone was when the program that a computer runs was moved to where the data is stored, rather than being represented or input physically. This effectively created what we now know of as software. Obvious in hindsight, yet almost impossible to see from the past’s vantage point.
Good way to describe ANPD.
on the same level as the document ‘contents’