Let's consider utf8 or latin1 encoding - german. When you look at the wording, you may notice that this question makes no sense for UTF-8 compilers with their default settings (these compilers would correctly produce an error while processing \usepackage[latin1]{inputenc}), so there is an implicit assumption that latex, pslatex or pdflatex are used, as opposed to lualatex or xelatex. However, the answer (with which you may agree or not) "use lualatex" has been accepted.

So, the OPer is ready to sacrifice the underlying assumption, and, thus, the question itself in its current formulation. Based on the accepted answer, my best guess is that the intended question was "Would you use latin1 or utf8 for creating German LaTeX documents?" (Btw., I feel that "use UTF-8 and compile with xelatex or lualatex" are pretty good general answers to the intended question, but this is off-topic here.) However, originally I read the OP as "Would you use latin1 or utf8 for creating German LaTeX documents if you use the package inputenc?" The implicit assumption "if you use the package inputenc" is IMHO equivalent to using latex, pslatex or pdflatex. For my own documents, the two questions (the intended one and the way I read it) would be vastly different; the OP misled me (which is reflected in my answer). Moreover, the question might be considered outdated, as @AndrewSwann says (the OP is old and the defaults have changed), cf. What shall we do with future out-of-date answers?.

I feel the community should EDIT the OP utf8 or latin1 encoding - german (making it clearer and more narrow in one or the other way or marking it as out of date), or DELETE it, or CLOSE it rather than leave it there in its misleading form. Of course, I left a comment to the OP.

Now, I don't feel senior enough to execute such a change or even ask the admin to do it without prior asking here, so, opinions are welcome: what to do with the OP utf8 or latin1 encoding - german? More generally, what to do with a broad, potentially misleading questions with several interpretations such that only the acceptance of an answer shows that the intended interpretation is notably different from the wording used?

  • That's a noble goal. However, only the OP can accept an answer, and if the OP accepted an answer usually the assumption is that this is what they wanted. As for the goal to improve the wording of the questions (in the sense of making them more precise): I really wish you will succeed, but I am not sure that there is a viable way.
    – user121799
    Commented Feb 4, 2019 at 2:20
  • Note that the original question is from 2010 and the answer marked as accepted the same year. Much has changed since then...in particular pdflatex was the standard choice and there was no utf8 default. So it is an out-of-date question, rather than an out-of-date answer cf. tex.meta.stackexchange.com/q/1878/15925 Commented Feb 4, 2019 at 10:28

4 Answers 4


While the general question in the title could have some answers or policy decisions, I don't think the example you give is an example of the problem.

I think you start out with an incorrect premise.

Let's consider utf8 or latin1 encoding - german. This question makes no sense for UTF-8 compilers

Firstly there is a natural interpretation of this question being "should I use UTF-8 or Latin 1 for German" You can't expect a question to get all the tex details right, so you always have to interpret tex fragments in questions with a certain amount of good will.

Secondly even if you want to strictly interpret it as a question about inputenc, then it still makes sense to ask about inputenc use with luatex or xetex. inputenc may not be that useful with Unicode engines, but we see examples posted here all the time that use inputenc with luatex or xetex. In the case of [utf8] it is harmless enough since it does nothing.

It is also possible to configure both luatex or xetex to not use their built in utf-8 decoder and pass the raw bytes to the macro layer as characters, in which case inputenc would work as in pdftex.

Finally while the collection of answers as a whole is useful as a resource for the general community, the choice of which answer to accept is purely the choice of the OP. If someone uses a crystal ball and answers the OPs real problem irrespective of the words in the question, and they get a tick of acceptance, that is just how it is, it isn't a problem that requires a solution.

  • I cleaned up comments as the site prompts: "Please avoid extended discussions in comments" Commented Feb 5, 2019 at 22:42
  • 1
    @user0 the listings answer was quite reasonable, the other one less so. But posting either on an ancient and only partially related question seems strange, especially when tied to this meta question querying the question, Commented Feb 5, 2019 at 22:49

If an OP accepts an answer, this means that their problem got solved by the answer. The decision whether or not an answer really answers the question can only be made by the OP asking it.

If you feel that this answer, nor any other answer given to this question, does not solve the problem described in the question, you can always ask a new question. In this new question you can link the related question(s), and explain why the (accepted) answers there do not solve your problem.

Questions do get edited, and usually it is checked that the OP agrees with the edit, or, at least does not object it. The question of whether or not one should edit a question to fit an existing answer has been discussed, and judging from the votes I'd say that the consensus is a clear "No".

If you feel a question is not clear enough, you can ask for clarification in a comment.

The model of this site is that votes will help to decide what the best or most appropriate answer is. That means that asking clear questions is supposed get rewarded. Every user has the right to judge themselves whether or not a question deserves an upvote. Whether or not these rewards really help to make the questions clearer I do not know.

From all the above, my answer to "What to do with questions whose accepted answer invalidates the question?" is that, if you feel the question is not clear, you may just refrain from upvoting it.


This is a common problem since it boils down to a fundamental problem of asking a good question. Not just asking a question, but asking a good, well-formulated, clear, concise, researched question.

The community attempts to guide the questioner if it is unclear through comments or sometimes answers. And answers are sometimes edited multiple times as the questioner struggles with their ultimate goal. The system is set up well enough to deal with these oddities, but ultimately you're dealing with humans. Most notable amongst these are X-Y type problems (also known as communication problems) where the main focus is a solution rather than addressing the OP's problem. Suggestions to the contrary or alternatives are often proposed in answers and it may be accepted without any consideration to the community or future visitors... that's just it.

If the question can be changed (edited) with minor changes to the original request in such a way that it better-reflects the question-answer connection, then go ahead and (suggest an) edit.

If the change would be substantial in nature and you have

  1. knowledge over the subject matter being discussed, and

  2. feel that the accepted answer does not address the aim of the question, while

  3. possessing the close vote privilege

then you should vote-to-close as being unclear. It can always be opened again if the OP rephrases the question to make more sense given the accepted answer. If you lack any of the above, consider discussing the issue in chat (where seasoned veterans and moderators readily roam) or coming to Meta.TeX.SE (like you did).

  • Here is an example of an X-Y problem where the OP just went with a solution provided, although the actual problem was never addressed.
    – Werner Mod
    Commented Feb 4, 2019 at 16:13

Expanding a little on my comment, note that the original question is from 2010 and the answer is marked as accepted the same year. Much has changed since then...in particular pdflatex was the standard choice, there was no utf8 default for LaTeX and as Caramdir's answer says lualatex and xelatex were still "experimental"/in development. Given that context, I don't think the question formulation was too bad. However, now for a new user of a modern latex it can indeed be unclear.

tex.meta.stackexchange.com/q/1878/15925 discusses the issue of out-of-date answers, but what we have here is perhaps an out-of-date question. I don't think this warrants closing though. The answer of Caramdir is good, but old, and strongly upvoted. A new answer could be added, pointing out the current situation. It could be made more prominent if awarded a bounty. As already discussed the accepted tick can only be changed by the OP, who I guess is no longer active on the site; but that is not a valid reason to close the question.


You must log in to answer this question.