Can we have a ruling on LaTeX on Stack Overflow?

Question

(Note, this isn't a duplicate of Where should a question about LaTeX usage go?, as the answers focussed on a specific question asked, which was about how to format bibliographies in LaTeX.)

It seems that LaTeX questions should go on Stack Overflow, not Super User. Reasons:

• nearly 300 questions with a LaTeX tag on Stack Overflow
• LaTeX is a combination of of document preparation and programming.
• Advanced LaTeX questions are not going to be answerable by non-programmers

Yet, Recommended build system for latex? (asked by me) got a vote-to-close (as well as other questions I've seen).

Can we get a ruling for general LaTeX questions?

EDIT:

Everyone here seems to agree. Can we get an argument from the people who vote to close?

Answer 1

Late addition: 1 September 2010:

With the Tex, LaTeX and Friends Stack Exchange beta site humming along and looking like it will survive (though we still haven't been appraised of the rules) I endorse posting most LaTeX questions there instead of to Stack Overflow.

I still stand by my position that such questions are on topic on SO (as outlined below), but feel you are likely to get a faster response on the specialty site.

---

I've been fighting this battle for months now.

LaTeX questions are appropriate on StackOverflow because:

• TeX and all its derived languages are Turing complete. Every document is a program, and problems are equivalent to "How do I get output X?" questions in any other language.
• Many problems are solved by "redefine macro \foo" or "define macro \Foo that extends \foo". These are programming solutions to programming problems.
• Some problems are solved by "get package Bar from CTAN", but if these are out than so are perl questions where the answer is "get Bar from CPAN", or even C++ questions where the answer is use Boost::Bar.
• If "it's a layout language" is an acceptable argument against LaTeX, it applies even more to HTML and CSS, both of which are widely accepted and much more prevalent on StackOverflow.

Moreover, at this point we have a increasingly nice collection (more than 250) of LaTeX questions and answers on StackOverflow, and more than a dozen users with 20 or more upvotes in the tag. Some of the questions are very basic, many are of fairly general application, and a few are really deep magic. Any attempt to more the focus to SuperUser would either fracture than population and expertise, or necessarily introduce questions to SuperUser which are rather more "programmy" than the culture there. The more esoteric ones would be even more out of place on SU than "How to I get boldface in LaTeX?" questions would be on SO.

---

Pure politics (rather than technical argument):

I know of only one LaTeX questions that has ever been closed on SO, and it didn't stay that way. This suggests to me that the pro-close side is rather smaller than the group of active users in the tag. My view would be that this in and of itself constitutes a consensus in favor of LaTeX on StackOverflow. YMMV.

Answer 2

First of all, I have nothing against HTML or CSS questions on stackoverflow.

That said, whenever I see a LaTeX question closed and I remember that there are several thousand HTML questions I don't understand the world anymore.

Tagged

• HTML: 5,737
• CSS: 4,232
• LaTeX: 272

What have questions like

What is the best way to create rounded corners using CSS?

or

What are good resources for CSS templates or templated layout sites?

to do with programming? And there are literally thousands of them on stackoverflow.

I'm perfectly OK with that, but please: relax when you see a LaTeX question.

Answer 3

I've changed my mind. I think dmckee is right, to paraphrase Latex questions might be better on TL&F, but they're just as OK on SO as they always were. But for the record, my original argument.

Now that tex.sx has launched, this question maybe deserves reconsideration. Suggestion:

1. The most important virtue of the new site is that it makes the whole SO/SU thing go away: the site is as happy with installation issues and pointers to packages as it is with nuances of \expandafter. So now, it makes much more sense in nearly all cases for questions to be asked on tex.sx, not SO or SU. 2. But there are two classes of exception. Firstly, there are questions whose trickiness is almost entirely in the system underneath: compiling LyX comes to mind as an example. These questions are quite likely to be better off on SU. Note, however, that anything to do with installing Texlive, Miktex, Mactex, or standard free PS/PDF tools are not going to be outside the competence of the tex.sx community. 3. Then there are questions that involve Tex, but are mostly about coding in, say Java: say putting together an Eclipse workflow, most of the problem being how to process the output of Pdftex in i-Text. These questions are likely to be better off on SO. But note, the tex.sx site has strengths in terms of coding both in C and in Lua, and in XML processing.

But now, I think the consensus should move to say that few Latex questions are appropriate for SO.

FWIW, I've made a Stack Exchange tag set tracking questions about Latex &c that are not on tex.sx.

Answer 4

LaTeX related questions on Superuser appear to be focused on integrating LaTeX with editors, identifying LaTeX files, and converting non-LaTeX files (e.g. Word) to LaTeX format.

I think that the general consensus of question submitters is that "How do I do X in LaTeX" is a programming question that's more appropriate for Stack Overflow.

Answer 5

Late to the party, but I now take a narrow view of what Latex questions should be allowed, and since dmckee's answer is the currently-accepted and highest-rated one, and since it's particularly terrible, I'll make my case by rebutting him.

TeX and all its derived languages are Turing complete. Every document is a program, and problems are equivalent to "How do I get output X?" questions in any other language.

This reasoning disingenuously overstates the sophistication of the typical Latex question on S.O. The fact that you can treat Tex like a Turing machine is, for a vast majority of questions on S.O. (and elsewhere), nothing more than a nice bit of trivia that factors rarely into the solution.

Now, if someone does need the programmatic power of Tex, such as if they're getting bugs while writing an original package, then that's a programming question, because they're actually programming. But it's fallacious to say that because a handful of Latex questions involve T.C.-ness and are programming-related, then any Latex question would involve T.C.-ness and be programming related.

Many problems are solved by "redefine macro \foo" or "define macro \Foo that extends \foo". These are programming solutions to programming problems.

Simple transformation macros could go either way, probably more in favor than against. Even so, it's the same sort of fallacy you committed in your previous point to think that this extends to any arbitrary question.

Some problems are solved by "get package Bar from CTAN", but if these are out than so are perl questions where the answer is "get Bar from CPAN", or even C++ questions where the answer is use Boost::Bar.

First, "some problems?" Try "most." But that's debatable unless someone wants to perform a census. But, that's not what's the best part. You see, Perl questions where the answer is "get Bar from CPAN" are justified because it's established that Perl is a programming language. C++ questions where the answer is "use Boost::Bar" are justified because it's established that C++ is a programming language.

You're arguing, then, that Latex questions where the answer is "use this Latex package" are justified because it's established that... wait, what was that you were trying to prove, again?

If "it's a layout language" is an acceptable argument against LaTeX, it applies even more to HTML and CSS, both of which are widely accepted and much more prevalent on StackOverflow.

I agree with your statement that Latex questions are comparable to another class of questions that are a clumsy fit for S.O.? Or were you doing the whole "two wrongs make a right"-style of justification. In any case, perhaps I should point out a certain site called Doctype.com, which is meant for HTML and CSS Q&A? Mayhap you've noticed the link to the site in the page footer on every Trilogy site? Mayhap you saw the same getting advertised on the Official Stack Overflow Blog?

While we certainly encourage web design questions involving HTML and CSS on Stack Overflow, it’s intended first and foremost for programmers. So if you’re a web designer, and less of a programmer, you might find a better audience for your question (plus some cool auto-screenshot capabilities) at Doctype.

What does this mean in practical terms?

• Our /faq will direct designers with pure HTML/CSS web design questions to Doctype, to better serve this subset of our audience.
• Our “hero roster” at the bottom of every page will offically list doctype.com as a member of the team.
• We will host Doctype discussion on meta, so the Doctype developers can participate and respond to community reactions and feedback for their site.

Moreover, I do not see any hypocrisy in stating that "O.K. guys, from this point on, if you're not making Tetris in Latex, your question's getting locked," and simply grandfathering in all the existing Latex questions. I do have a problem with using inertia as a reason to maintain the status quo.

Also, with the passage of time it now seems fairly likely that Latex and co. will get their own dedicated S.E. site, which weakens the argument that Latex is allowed on S.O. even further.