My usual operandi is: poke around or Google for a solution, until I figure out one of the fifty ways that achieves my desired outcome. I then declare my problem solved. If my solution is deprecated (by whom?), then whoever deprecated it may tsk-tsk to their heart's content.
Most mathematicians I know are not all that fussy with how they write their LaTeX documents, the same way programmers can be extremely fussy with their code. If there are "best practices" out there, then they aren't widely known or followed.
That said, here's something that may interest you. On the arXiv (arxiv.org), at least in math and probably in other disciplines as well, the LaTeX code is available for nearly any paper you want. Pick your favorite paper, and then click "Download other formats" on the right.
If nothing else, you can see what professional researchers are using in practice.
Well in the simplest case how about something like \bf versus \bfseries versus \textbf?
Google that and you'll find thousands of blog posts, each telling you which is the One True Way™ but not expanding on the why.
Finding actual answers is really hard for latex.
Use any of them.
> thousands of blog posts, each telling you which is the One True Way™
When I don't know how to do something in LaTeX, here is my flowchart for figuring out how:
1. Google
2. Look at any of the thousands of blog posts
3. Try the proposed solution, while ignoring any "One True Way" crap
4. Are the results what I wanted? If yes then done; if no go to Step 1.
Now LaTeX technically is a programming language, it's Turing complete. But in practice for most users it's not.
Sure, there might be subtle inconsistencies. There might be some bit of mathematical notation that looks slightly different in one part of your document than another. I guess some people would be bothered by this, I'm not one of them.
It is true that it's possible to become a LaTeX power user, and write all sorts of sophisticated macros. In that case, yeah, you want to get the foundations right. But if you're just doing something simple (e.g. writing up a math paper), then mysterious breakages aren't an issue; there's not really anything to break.
When I worked as a programmer in industry, I was quite anal retentive about how I coded, and getting a feel for the languages I used. They were the tools with which I did my work. But LaTeX, for me anyway, is merely the tool with which I typeset my work. An outstanding tool, perfect for the job -- but not one I choose to invest time in mastering.
It has often been repeated that we should all aim toward separation of content and presentation and LaTeX has been highly touted in this area, yet I find it very difficult to achieve in practice. Far too often, I find some package-provided environment lacking and I need to manually add some extra space or what have you. Now my document has become tainted due to the introduction of explicit styling in the content file. It's now baked in at that location and is very difficult to update later.
How do I avoid doing this? It's not enough to say "go back and redefine that command/environment to make it work the way you want." LaTeX is a leaky abstraction.
When writing my thesis I used LyX but found myself creating more and more ERT (what LyX calls raw LaTeX code). I eventually just started using a text editor. However, I still do tables in LyX and copy them over -- such a pain to insert a column otherwise...
At that point why not just use .docx?
Complex formatting, such as figure captions and figure placement, in Word documents suddenly break for no apparent reason. Troubleshooting a “broken” Word doc is akin to fixing a typewriter while one is blindfolded.
Yes. In 1994, you would `latex document` and then `dvips document` to get your document.ps. Then in the 00s almost everyone would use `pdflatex document`. And these days, pdflatex is going out of fashion, but I am unsure what the current to go tool is.
(This just runs latex and bibtex in the right order and the right number of times..)
Unless the authors managed to solve the halting problem, I guarantee that latexmk doesn't always run them the right number of times. ;-)
You can disable the maximum number of runs, in which case it really does run the right number of times.
I'd pay good money for a "modern idiomatic latex"-type of tutorial.