Shell Script RSS Publishing | Hacker News Reader

Shell Script RSS Publishing

shinobi.bt.ht

89 pointsFred3416 comments

It's telling the the landing page is actually HTML so links fucking work. If you want to go super minimal with "plain text", which has a variety of meanings, go ahead. Minimalism is fine. I just find the "plain text" fetishism silly.

The web won out over FTP and Gopher because documents could link to others for reference. Those systems could only link through menus (if at all).

HTML is actually pretty awesome. If you want minimal HTML just make minimum HTML. You don't need JavaScript or CSS. You can include a single meta tag with a viewport directive and get a magically responsive page perfectly readable between desktop and mobile without having hard line breaks.

Plain text on the web sucks. If the browser gets a header of "Content-Type: text/plain" what's the encoding? Good luck figuring it out. While ISO-8859-1 is defined as the default for text/plain I really hope you don't use any high ASCII characters or a European keyboard. A lot of web servers won't automatically add a charset to the content type. With HTML you can define the document's charset in the header of the document itself. Plain text also has no semantic elements so the browser can't make good formatting decisions in absence of explicit formatting instructions.

I think hypertext is probably one of the most underappreciated technological advancements. You can do some incredibly cool things if you really lean into it.

Ted Nelson was right all along.

Yeah it's definitely something we take for granted now. It's changed our brains.

I used network services without hypertext, in the form of BBSes, and it seems like worlds away. I guess we just described in English how to find things online back then, and didn't link to them? I think there were FAQs of the era that were like that?

Would be interesting to read early FAQs on Usenet, pre-web (anyone have an archive? [1])

It's funny that there used to be entire hypertext conferences back in the 80's. I guess that field is "dead" simply because it's been absorbed into standard practice? (the web)

Then again Nelson's hypertext had all sorts of bells and whistles. I'm tempted to say that the web gives you the 90% of the functionality that actually scales world wide, for 10% of the cost

Though certainly there could have been better outcomes

[1] Hm not sure if there are any pre-web ones here - http://www.faqs.org/faqs/by-category.html

hypertext is "dead" because we learned to stop worrying and love dead links.

(VNC could mirror not only real displays but also virtual ones. imagine if it had had hyperlinks between the latter?)

You are right that we should not got back from html. I find .txt file publishing interesting anyhow because the files can be rendered in everything. A .txt file can be viewed in a browser, rss reader, any text editor, in the terminal, etc. .With a html document you are limiting yourself to viewing the information in a browser or an email/rss client the supports html rendering. There is more redundancy with publishing through .txt files

HTML isn't some arcane binary format or compressed or something. You can view it anywhere you can view a text file. If as a publisher you want your HTML to be readable in a text viewer it's trivial to run tidy (or whatever linter you prefer) on it. You'll get nicely indented text blocks perfectly readable in a text viewer. The output is no less readable than "plain text" marked up with a bunch of faux page layout like hard line breaks and lines of dashes under headings and the like.

The JavaScripter über alles crowd generates some downright abusive HTML and CSS even when generated as a static document but there's no intrinsic reason for that. HTML can be a fairly lightweight markup on top of text. Even while being lightweight it can bring significant enhancement over plain text by being able to carry metadata, encoding information, semantic information, and even provenance through comment tags.

On the character encoding I think there are just people who care and people who don’t.

Those who don’t will mess it up regardless of mime type. Those who do will take the two minutes to set it up whether this requires a server configured header or not.

(Agree this is a slightly annoying affectation.)

The client could always detect URIs in plain text to make them clickable. As a human-readable alternative, there’s also Markdown or Gemtext.

I guess could also have a markup format that let you turn an arbitrary bit of text to a link to a URI. That markup could itself just be type-able plain text characters so the whole document was plain text at a low level.

Oh...

HTML degrades to plain text less gracefully than Markdown. In general, it is also far more complex.

Shameless plug: shell script static site generator https://mkws.sh/

I love it. I've wanted something like this ever since nanoblogger died. Cheers.

Thanks! Check out the themes also: https://t.mkws.sh/

You can replace the bin/mkws script with a Plan9 mk file to get incremental builds:

  #!/usr/local/plan9/bin/mk -f

  URL=//example.com
  SHARE=share
  LANG=en_US.UTF-8

  THEMEFILES=$SHARE/l.upphtml\
          $SHARE/s.uppcss\

  TPLFILES=`{find . ! -name "l.upphtml" -name "*.upphtml" | cut -c3-}
  HTMLFILES=${TPLFILES:%.upphtml=%.html}

  $URL:V: sitemap.xml $HTMLFILES

  %.html:Q: %.upphtml $THEMEFILES
          >&2 echo Making $target
          pp $SHARE/l.upphtml ${target%.html}.upphtml $url > $target

  sitemap.xml:Q: $HTMLFILES $SHARE/sitemap.uppxml
          >&2 echo Making $target
          pp $SHARE/sitemap.uppxml $url > $target

  clean:VQ:
          Removing "$HTMLFILES" sitemap.xml
          rm -f $HTMLFILES sitemap.xml

It's drop in!

Checking in on the site from a year ago that claimed to be using this, and the discussion then:

https://news.ycombinator.com/item?id=31372373

RSS is just an XML document; you can generate it and serve it with CGI using language.

s/using language/using any language/

Reply on news.ycombinator.com