Show HN: Plain text for the web
saola.in
saola.in
In 91 or so it was taking advantage of the macs flat address space to do rpc between itself and apps (I wrote a module that let you receive events from the finder).
Around 94, Dave (Winer, the author) was using to handle cgi scripts in "usertalk" its native language, from WebStar, a Mac web server. People would set up web based workflow to (then Mac only) apps like FileMaker and debabelizer.
A few years after that, Dave realized he could be the server himself, and created an app that was basically a webserver living on localhost and let you edit locally and publish your weblogs to a static host.
So, not the same because it's not flat files, but with similar possibilities.
(And a call out to Macintosh Programmers Workshop, which was UNIX like, but allowed selections within arbitrary files be treated as first class files - you could execute a command to take select portion of window 1, pass it to the compiler a land redirect error messages to a section of which now 2. Many a build system came to life this way... Sorry nostalgia...)
1. What's different about this setup? What can I do that I either can't with a standard webserver setup or would be more cumbersome that way? 2. Why "files within files", what's different about that compared to directories and index files?What can I do that I either can't with a standard directory structure or would be more cumbersome that way?
If I wanted plain text at the moment + scripts, any standard webserver + CGI would give me that right now and I could host it immediately on a large number of hosts (or on my own). So for people like me who would go "ooh, plain text and unix sounds nice" it'd be good to see a set of reasons why what we'd currently do would be improved using this.
Again a webserver+CGI would definitely work no doubt. The app just simplifies it for you. Its the same thing with content management system. Its simpler to use them rather than build it manually. Also I plan to add the ability to share files.
You allow people to execute scripts on your server? How does the security aspect of that work?
I prefer human readable formats since it avoids lock in and allows me to edit my content on any platform. The downside of course is lack of media support. Quiver is doing some interesting stuff in that area, but at the cost of losing some human readability.
1: nvALT http://brettterpstra.com/projects/nvalt/
2: TaskPaper http://www.hogbaysoftware.com/products/taskpaper
3: Quiver http://happenapps.com/#quiver
https://www.madoko.net/ I've been using this for ages, it's a superset of Markdown, pretty solid. It's really hard to beat (La|Te|Lua)TeX for certain things (I'd never write a dissertation or master's thesis on Word because even with all of the plug-ins it's absolutely abysmal for citation management (BibTex) and rendering even the most trivial of diff eqs) but this gives TeX a run for its money. It's a pretty decent option which seems like it might be up your alley. (Org-mode is probably what you're looking for though).
It's not totally full-featured yet but it has some decent support so far.
So you can have local scripts that affect the files and the changes will be synced for your other devices to see.
I use google keep for online plain text type things. I can definitely see the power of being able to run scripts over these kinds of notes.
For years I had been using a system which was built around yahoo pipes and cron tasks. Pipes would suck in rss sources from a huge number of places, do some munging and logic and output another rss feed which I would process with a cron task to do various things based on what was in the feed.
Then yahoo killed pipes.
Can a script create a static output file?
The whole idea was that sites I use for things come and go. I could jump from service to service, update my pipes setup. If a site died I'd just find a replacement and move on. I always had all my content harvested and stored in my own db.
Examples;
* any youtube video I added to my favorites would be auto embedded on my blog
* posts on blogger, wordpress, etc all pull into the one place
* images fav'd on deviantart
* pages book marked using diigo
* photos saved to whatever photo site I was currently using
Here's a screenshot from a long time ago when I originally set it up (2009):
https://fuzzyslogicblog.files.wordpress.com/2009/03/a186d-pi...
I imagine a few "standard scripts" like your calculator might be useful to show some of the things you could do and might increase a initial user's stickiness.
> "In the spirit of UNIX, every thing is a file." "Everything is a file" is a Plan 9, not UNIX, adage. `Block devices aren't a file (though they are represented by files). Network devices are not files (again, they have representational forms, but you can't pipe to them, a la Plan 9). > "Though unlike it, a file can contain other files. " By definition, a directory is a file. It's the second item in that tuple of 3 outlined by Ritchie.
You essentially have re-invented Plan 9 and to some extent ACME[1]. Sadly, Plan 9 faded into obscurity due to licensing, but it was poised to fill the position Linux has now. Watch that video [the whole thing) and borrow the good ideas. LISPs are homoiconic in that whole text-is-data sense. ACME takes it a further, in that your terminal is a shell, which can invoke programs to manipulate data. The only other environment I've ever worked where everything was as unified consistently was Smalltalk.
Emacs is a close 3rd and that's hard for me to admit, as an emacs fanatic. Though I can (and have been) doing exactly what you're trying to solve for ages within emacs, but that leaves all the vimmers in the dark ;) Ledger for accounting, org-mode for generic notes, a highly customized e-mail client which I integrated to a calendar/contacts/todo, all of which have tags. I can fuzzy-match my notes to the point where if I'm trying to remember a resource (say, a library for redistribution of data amongst a network on soft-failures), I can search by tag, by time (i.e., I remember it was sometime this week; filter; I remember it was a link on medium.com; filter), and various other components.
I see what target demographic you're going for and it's a pretty good idea. Most people need 'information stores'. Evernote was the Web 1.5/2.0 solution, inevitably there'll be a Web 3.0 version, you might be it.
Side bit of trivia: Cox loved ACME so much he ported all of the essentials when he moved over to Google, from the Plan9 FS (FUSE'd, not native, obviously) to ACME itself.
[0] http://www.tuhs.org/Archive/PDP-11/Distributions/research/Mc... [1] https://www.youtube.com/watch?v=dP1xVpMPn8M
Speaking of standards and processing it would be cool to have a FUSE interface such that I could hand edit ledger files or org mode files on the thing, then via a FUSE mount, continue processing or accessing stuff on my desktop or a server.
Sorry if correcting a typo is not an appropriate comment.
I have used Google Drive and Google Apps scripting for this kind of thing in the past, so I'd be interested to know how this compares.
Cheers