I’ve wanted to learn the roff ecosystem since I learned W. Richard Stevens (RIP) used troff for his books. 25 years later it’s probably time to accept the fact that I won’t.
Given that we have latex, pandoc, markdown, asciidoc, etc., I really don't see the need to learn it. At the time it was probably a great and transformative tool, but using TeXMaker or TeXStudio and latex gives a far more comfortable writing environment and the documents you can make are much better looking.
I wouldn't manually write documents in groff these days, but I have found that it can still be useful for automatically generated PDF reports/documents.
Groff is lightweight and already installed on most Linux systems, and the build process is way cleaner than TeX. It's easy to make a Python script that reads some data and prints a nice-looking report (or a form letter, etc.)
Yes, integrating it into a scripted workflow, including dependency-based build systems, is vastly easier. The multiple runs of TeX don't fit well into this model, and the Unix heritage of troff vs the TeX is clearly apparent in the design, which permits all the bits to be piped together. It's clearly dated and limited in many ways, but it's still extremely powerful, and it can do automated document generation very well. I've used it for this in the past, and would recommend it.
While we do have more modern stuff like Markdown and ReStructured Text, these are in their own way extremely limited and are primarily focussed upon transformation to HTML. troff is focussed upon printing, and if you want nice PDF reports, and you don't want the slowness and excessive disk space consumption of TeX, it can be an excellent solution.
> still extremely powerful, and it can do automated document
generation very well.
These days I am more likely to be working in org-mode and using elisp
snippets to automate stuff, but roff is a solid fallback for things
like quickly turning 100 paragraphs of individual feedback into 100
formatted pdf documents that can be mailed out.