It's been a little more than two weeks since I first published my inaugural "writing report," which is to say the output of a simple shell script I concocted using find, cat and wc to count up the number of words I have written in a given period. Currently, I am sampling in the 14 days range, as this eliminates some of the skew implied by days where I write little, or days where I write much. I haven't really tested this, statistically: it's just a hunch.

In the interim between the first installment and this one, I have made a number of refinements and changes to my basic workflow. The major one will be evident from the output of the shell script alone: I have moved away from word processors entirely. This isn't some free software gimmick (I use LibreOffice anyway, when I need that kind of tooling):– it's more of an attempt to learn as much as I can about Emacs and leverage its power.

As a former, "retired" software developer, I am trying to enjoy programming, and I kind of like Emacs Lisp in a weird way. That's by the bye, though.

In any case, the report gives some indication of the favor being shown to Org-mode at the moment, with some work done here and there in LaTeX. Even the latter, however, is often done through Org's export feature, and including it in this metric, "as-is," has the moderate downside that the script naively counts all the LaTeX directives as words. Something I will probably change in the future, though for now, it does not bother me too much.

More exciting by far is my discovering how to better use Org's plain text buffers for doing meaningful, multi-phase edits to documents. One of the reasons I and many other people rely on word processors is for their "track changes" feature. In a WSIWYG editor like LibreOffice Writer, it is an excellent supplement to editing prose. The ability to add comments in the document gutter is great, too. But, because it's not a raw view of the document, as it exists on disk, it obscures the fact that ODT documents are, like their proprietary counterparts, ZIP archives containing a bunch of XML files. This not only makes them hard to search, when you are using tools like grep, but it also makes looking at just comments or just edits done in track changes, a pain, without firing up the application.

Emacs' built-in alternative is undoubtedly more primitive, but, with some reservations, I would say it's almost as good. What preserves its utility, chez moi, is the ability to keep everything in one two document(s). You navigate to your buffer of choice and active highlight-changes-mode, which prompts you to select another buffer to compare. This necessitates copying stuff, but it's a workflow I like. I start taking notes and doing some basic drafting in an Org file that has some tree structure to it, then, when some volume of prose begins to accumulate, I copy it over to another, fresh buffer, and start doing my edits, using a combination of Org's strikethrough overlay, and the aforementioned minor mode.

That's all for this week, tune in again whenever the next one comes!