[nextos]

I just logged in to say this. It's an incredibly powerful feature.

In general, plain text is great because it's explicit. In contrast, word processors store some odd chain of objects. The steps to rebuild a document from scratch that leads to the same representation are not totally obvious. Sometimes it's a bit stochastic even!

XML formats improved the situation, but still plain text wins IMHO. With plain text you can have a toolchain where components for version control, edition and transformation can be replaced seamlessly. New stuff can be introduced at will. In e.g. Word you're mostly locked-in, or at least it's far from trivial to break out.

[wcaleb]

Original author here. Totally agree about version control. When I wrote that post I had not yet learned how to use Git. In a follow-up post I discuss how I now manage all of my notes and citations in plain-text on a Pandoc- and git-powered wiki called Gitit.

http://wcm1.web.rice.edu/plain-text-citations.html

[kjax]

I appreciate that you mentioned LaTeX and its potential usefulness for these scenarios. You get the ability to version control, without making things especially difficult to translate your text into a more appealing output. Granted, one has to learn typesetting semantics first, but it seems to be a good middle ground.

[explosion]

wcaleb, this is great, thanks for both posts. It's also very inspiring to see your open notebook workflow and experiments.

[nextos]

There's a very timely HN post discussing a vim implementation on top of emacs: https://news.ycombinator.com/item?id=9394144

As much as I like vi's interface, I think emacs implementation is superior, and org-mode invaluable.

[nextos]

Looks great, just watched the video. Have you considered org-mode?

[chriswarbo]

I tried org-mode for a while, but found it overly complicated. I use Emacs heavily, but developed a couple of Pandoc extensions which give me the org-mode features I need without all of the baggage: http://chriswarbo.net/essays/activecode/index.html

[wcaleb]

Have looked at, but not really considered. I went down the Vim path pretty early on ...

[peatmoss]

You can use org-mode with Vim more or less the same as you'd use Markdown if you're planning to compile your doc with Pandoc. You can even use README.org on github in place of README.md, and it will be turned into HTML automagically the same way as Markdown.

Why use org-mode as a format instead of Markdown? Internal cross-references. Org-mode feels just as light weight to me as Markdown for the most part, but it has fewer cases where you can't make things work. Worst case in org-mode, you mix in LaTeX.

[CyberpunkDad]

org-mode is decent for LaTex markdown substitution. * = Section = SubSection

Etc.

I think I still prefer to just write it in markdown. Word can't come close to the typesetting ability of LaTex.

[marvel_boy]

Looks great, thanks a lot for sharing your workflow !

[the_af]

Agreed. This is the same reason why some of us are skeptical of the frequent calls for more "visual" programming languages that eschew textual representations of code.

[digi_owl]

Thinking about block based "IDEs"?

I think the biggest difference there is that a document is "static". It is not supposed to process something else like code is.

Thus what you want to avoid at all costs is "boiler plate" bugs. Bugs that are not caught at compile time because it is semantically correct, but that end up producing wild results once a certain kind of data is introduced.

[hackuser]

> In contrast, word processors store some odd chain of objects. The steps to rebuild a document from scratch that leads to the same representation are not totally obvious.

Microsoft Word's binary format (.doc) stored the document's text as a block in the file, making data recovery simple.

[digi_owl]

I seem to recall older word processors having a "explicit" mode that allowed the editing of markup directly. But i don't know if that is still present, much less used, in the age of "drag and drop".