For the longest time, I've been typing my prose into plain text files. That has a lot of advantages: they can be opened with pretty much anything, they're as compact as files get without compression, and can easily be turned into web pages or e-books through Markdown. Next time however I might just try to use WordGrinder instead.
Wait, what? WordGrinder (available from cowlark.com and various Linux distributions) is a word processor in the old sense of the term, from before humongous office suites became the norm: a program designed to let writers write, with as little fuss as possible. You get a word count (and paragraph count), formatting roughly on par with the aforementioned Markdown, a decent range of import and export options, and a spellchecker. That's it!
More importantly, you get all that from a program not one megabyte in size with all dependencies, that can run in terminal emulators (and X11). Talk about software you can install on toasters! For someone like me, who uses computers so ancient that even AbiWord has noticeable overhead, it's amazing.
Even better, WordGrinder has some unique and valuable traits. Also some quirks, but for once they're part of the charm here.
The biggest one is that WordGrinder supports multiple documents in a single file: an obvious thing to have, yet for some reason I can't think of any other except Sigil, which uses EPUB as its native format (but isn't nearly as good at simply letting you type, and can only import or export HTML).
Also, the user interface is minimal but not simplistic, and uses modern, familiar key combinations. Hitting Escape opens a small menu system you can then navigate with the arrow keys (but not the mouse), or with the shortcuts it shows you. There are pretty, visible indicators for the start and end of a document, and the cursor always stays in the middle of the window. That kind of wastes space and reduces the amount of context when you're writing as opposed to editing, but it makes sense and has charm. Which seems to matter.
Last but not least, WordGrinder's native file format is text-based and human-readable, so if you ever find yourself stuck with a bunch of .wg files and no ability to run the software that made them, it should be easy to recover the data. Well, in principle the same is true of LibreOffice. Whose .odt files WordGrinder can import just fine, by the way. It seemed to choke on HTML at first, until trying to export instead revealed the reason: you have to make it a complete XHTML document instead. In other words (ha!) the raw output from Markdown doesn't work, but a part from an EPUB file should.
All right, that got very technical very quickly, and I should wrap up this review anyway. Which by the way I'm writing in the software it's about, and it's working out very well indeed. So maybe give it a try if you can, yes?