9

I think it's important that the Questions and Answers on our site remain "on-topic": It is infinitely frustrating when trying to solve an issue, to search for a question, find an answer, and realize that the answer is invalid for Ubuntu.

I would like some clarification on what should be done about answers that explicitly say that they pertain to another operating system.

Per "Are answers specific to Debian on-topic?", the criteria for whether or not an answer should be posted seems to be that the answer must apply to Ubuntu, otherwise it's not a valid answer and should not be posted.

Answers about Debian are one thing; I think they are probably fine due to the close relationship between Debian and Ubuntu.

But what should we do about answers that explicitly apply to another operating system that is very dissimilar from Ubuntu, like Arch, Fedora, or SUSE? Answers pertaining to these distros are much less likely to apply to Ubuntu. For what reasons should we keep or reject these answers?

| |
8

TL;DR

As Ravexina says, I think this is about that answer, which appears intended to work on Ubuntu, but which discloses that it was tested on Fedora. Considering that it would not be considered a moderation issue if the answer had merely not mentioned it was tested at all, I don't support deleting that. What this comes down to is an issue of correctness, and as Kulfy says, that's what votes are for.

But I admit there is some support for your position (though not from the help page on what questions people can ask). It is good you've posted about this. I've been somewhat aware of this issue for a while, and I should've posted myself. I think deleting answers for mentioning they were tested on another distro is a very, very bad idea, and that a policy that all answer authors must test their answers on Ubuntu would support deleting an enormous amount of valuable, highly upvoted content on the site. But I also acknowledge that my view may be in the minority, and that this is not my call to make.

When answers are wrong

It is frustrating when an answer is invalid for Ubuntu.

Of course, most answers that are invalid for Ubuntu don't have anything to do with other operating systems, and don't mention any other operating systems. They're simply wrong answers. We get a lot of wrong answers. We get a lot of answers that are not tested, or not fully tested, or not tested under the specific conditions the OP had. Many of these are wrong answers. Many others are not wrong, and it is often not feasible to reproduce every aspect of a problem the OP had.

Ask Ubuntu, being a Stack Exchange site, is built from the ground up with a mechanism to address answers that are wrong, misleading, or otherwise low quality: downvotes. (It is also useful to comment and, when an answer can be improved without radically changing its meaning, to edit.) Upvotes are the main way to indicate that an answer is correct and useful, and downvotes are the main way to indicate that an answer is wrong or not useful.

Of course, you might consider it harmful or unfair to downvote an answer that could well be right and that you have no strong reason to think is wrong. But if that is the case, outright deleting the answer is far worse!

When answers don't even try to answer the question

We should interpret questions as being about Ubuntu when it is reasonable to do so, and we should close them when it is not. We should judge answers based, first, on whether they're even attempting to answer the question they're posted on, and second, whether they succeed in this goal. I think--and this is my opinion, not an established policy--that there is a big difference between:

  1. an answer that is intended only to work for some operating system other than Ubuntu and is thus not even an attempt to answer the question it was posted on.
  2. an answer that is intended to work on Ubuntu but discloses information that suggests it might not.

In my opinion, the first is NAA--though occasionally it could be fixed by editing--but the second is just fine. As a matter of moderation, we don't actually require that people test their answers at all. If someone has tested their answer on Ubuntu, that's good, and it speaks to the quality of their answer. If someone has tested it on some other operating system, that's less useful, and sometimes not useful at all, but it doesn't mean their answer is not intended to answer the question that was asked. Furthermore, it's a good thing when answers disclose possible shortcomings.

This site is based on the idea that we, together, are reasonably capable of judging whether answers to questions about Ubuntu are correct. This is something we're in a good position to do, because most of us use Ubuntu and have some knowledge of Ubuntu.

We're not in a good position to judge other operating systems, whether for similarity to Ubuntu or otherwise. For example, you've proposed allowing answers that mention Debian specifically. What about other Debian derivatives? What about unofficial Ubuntu derivatives? Furthermore, how similar or dissimilar Ubuntu is to another OS depends on what topic is being asked about. Fortunately, knowledge of other OSes is not needed to judge whether an answer is correct for Ubuntu.

...but our existing policy may actually support your position.

Considering that I am constantly going on about the importance of community consensus and the value in citing existing policy, it would be bad--like, real, real bad--if I failed to mention that, at least at the time of this writing, the existing community consensus on this issue appears to favor your position and not mine.

On the one hand, I think the answers to the meta question you cited here, Are answers specific to Debian on-topic?, actually support my view. They're about what OS an answer actually applies to, not merely whether or not it has been tested on Ubuntu at the time it was posted.

But on the other hand, as Ravexina pointed out, we have the more general question What should I do with answers about other Linux distros?. One answer (by TheWanderer) has decisively more votes than any others, and it basically supports your position on this. It begins by saying:

I tend to follow the same mentality I use for questions. Questions about Mint, Kali or even macOS could be perfectly applicable to Ubuntu, but I don't know that, and so I flag as OT. There are specific rules about this for questions, but I use this reasoning to explain my decisions if needed.

An answer that's based off of RHEL, Kali, Debian, etc could certainly work perfectly fine on Ubuntu, but there's no way to really know unless you or the OP tests it. For most cases, I'd flag the answer if it said something like "this worked on Fedora..."

Now, I really strongly disagree with this. Answers are not questions--to answer a question, we have to know about the system the OP is using, but to judge an answer, we have to know about Ubuntu. So a question about a non-Ubuntu OS is actually about a non-Ubuntu OS, while an answer tested on a non-Ubuntu OS may still be an attempt to answer a question about Ubuntu. Furthermore, the need to actually check if an answer works is in no way limited to answers that-- Well, it is in no way limited at all.

But perhaps this criticism is too harsh, considering how that answer then goes on to carve out a major exception:

However, there are cases where an answer might be all-round Linux applicable. If it's something like:

Q: How do I copy a file from a to b in terminal?

A: I use RHEL, and this is what works: sudo cp /file /dir/

Then the RHEL is pretty much irrelevant. Something that simple, it doesn't really matter what distro the answerer uses, since it's definitely applicable to Ubuntu.

Something more complex, like VNC/RDP, should probably only get answers from Ubuntu users.

It's challenging to apply this recommendation, because the one example it gives of something that's the same on all distros is a textbook case of something that is not the same on all distros. Not all distros ship with, or recommend, sudo. Furthermore, the example mentions CentOS. I haven't used CentOS in a while, but last time I used it, sudo quite deliberately did not work out of the box.

Deleting all answers by authors who aren't Ubuntu users is not something anyone seems to support. But deleting many such answers, while trying to protect others that seem obviously okay, appears very popular. In that answer, simplicity was the guideline. In what you've proposed, distro similarity is the guideline. But I think it is precisely that kind of judgements that go into such policymaking that we are ill-equipped to do here, and precisely the kind of judgments that go into commenting and voting on possibly wrong answers that we are well-equipped to do here.

I don't think I'm the only one uncomfortable with this. I believe that policy is rarely applied, and even more rarely cited. But that doesn't make it right for me to simply ignore it.

A disturbing trend

I see efforts to flag and delete answers that mention they were tested on another operating system as part of a larger trend of using flags and delete votes in place of downvotes. I'm not saying this is something you often do, but I am saying it's something that gets done a lot on this site.

It has become common that I see an answer I know to be correct that has been deleted on the grounds that it's clearly wrong. I often edit these answers to clarify them, and vote to undelete them (or flag them if deleted by a moderator, since then the system doesn't let me cast an undelete vote).

But I also just ignore a lot of them. I have to pick my battles. It takes a lot less time and energy to cast a delete vote or raise a flag, than it takes to edit a post and then explain why it should be undeleted. And that's just the ones I happen to know are correct.

I wish, at minimum, that people would stop trying to delete things they aren't even comfortable downvoting.

| |
  • 2
    I very much agree with the statement "I see efforts to flag and delete answers that mention they were tested on another operating system as part of a larger trend of using flags and delete votes in place of downvotes" - it should be the community that decides the answer 's worth rather than mods deleting just because grep matched a term. I posted a very organized, technically detailed, & mostly distro-agnostic (Ubuntu-based anyway) answer to this question last night, only to see it deleted for bc I mentioned testing it on another distro. frustrating – zpangwin Jul 9 at 1:23
  • 2
    There's a lot to think about here and I think you've won me over to your side. I agree that it is probably better in most cases to have an answer that might not work, rather than for it to not exist at all. Where I might still feel uncomfortable might be if the answer was a first answer, since unanswered questions have a higher priority. – Nmath Jul 9 at 4:19
  • 2
    @Nmath I'd say the case of a first answer is especially one where voting rather than flagging is the right tool. The system considers a question "answered" when it has at least one answer whose score is at least +1 or that the OP has marked as accepted. New answers that haven't been voted on yet, and any answer that doesn't have more upvotes than downvotes, do not cause a question to be considered answered by the system. (Note the descriptive text on the list of "unanswered" questions: "123,161 questions with no upvoted or accepted answers".) – Eliah Kagan Jul 9 at 5:59
  • IMHO, you've got the "disturbing trend" backwards. Answers only tested on other distros are flagged or downvoted as off topic, because they are not "about Ubuntu", according to the same logic that applies to questions. I've not seen an answer downvoted for just a mention of another distro. – mikewhatever Jul 10 at 8:50
2

If an answer isn't applicable to Ubuntu, then it's simply not an answer and should be flagged as such.

However...

  • If an answer is applicable to other distributions (doesn't matter whether they're Debian-based and/or similar to Ubuntu) as well as Ubuntu, it should be OK to post/keep relevant info.

  • Also if an answer contains addendum as parts of an answer, that are applicable to other distributions, that should be OK, too.
    For example, "To fix XYZ on Ubuntu, do ABC; For the same problem on RHEL, you'd instead do PQR". After all, it contains the necessary answer for Ubuntu, and we're not against other distros or sharing knowledge/information as long as there's a valid answer for Ubuntu.

With respect to this answer:

It's not clear whether it'd work on Ubuntu as well. So if somebody confirms that it works on Ubuntu as well, great, it's a valid answer.

Similarly, if somebody confirms it does not work on Ubuntu, then it should be flagged as a non-answer and probably deleted.

But as it is, it's an ambiguous situation. It may turn out to be relevant answer or may not. Under the circumstances, you did the right thing by pointing this out.


In general, an answer posted to any question is intrinsically tied to the question; doesn't exist own its own. Generic answer, additional information, historical artefacts or background, etc are fine. But it should be an answer to the question to start with.

The point of having a dedicated https://askubuntu.com site is to find solutions for problems pertaining to Ubuntu ecosystem. As noted previously, it's fine to post answers that may also applicable to other distros as long as it contains answer(s) valid for Ubuntu. Without that, it's just a useless noise. Otherwise, why not merge https://askubuntu.com with https://unix.stackexchange.com?

It may be relevant to refer to the history, too.

| |
  • 1
    I'm not sure what you find ridiculous about my comment. The help page about what questions people are allowed to ask here is about questions. It does not speak to the matter of what answers can be posted. This does not mean that it is okay to post anything at all as an answer, but when something is not okay to post an answer, that help page is still not the reason. When one cites a source to support a claim, the source should actually support the claim. – Eliah Kagan Jul 9 at 1:05
  • I may have misunderstood your comment - that you're citing the "what to ask page" doesn't mention anything about the answer, so it's OK to post possibly-unrelated answers. But it looks like you're in fact questioning the previous comment there in the same line I did here. – P.P Jul 9 at 8:01
  • I have removed that part as it's very much a misunderstanding of mine. I don't think we disagree in general on how to handle this or what's acceptable. But I am favour of deleting that answer if it turns out to be not applicable to Ubuntu. But as things stand, it doesn't warrant deletion or even downvote - simply because I don't know the validity of the answer. – P.P Jul 9 at 8:06
1

I largely agree with PP's observations but would expand his original list to include one other exception:

If an answer isn't applicable to Ubuntu, then it's simply not an answer and should be flagged as such.

However...

  1. If an answer is applicable to other distributions (doesn't matter whether they're Debian-based and/or similar to Ubuntu) as well as Ubuntu, it should be OK to post/keep relevant info.
  2. Also if an answer contains addendum as parts of an answer, that are applicable to other distributions, that should be OK, too. For example, "To fix XYZ on Ubuntu, do ABC; For the same problem on RHEL, you'd instead do PQR". After all, it contains the necessary answer for Ubuntu, and we're not against other distros or sharing knowledge/information as long as there's a valid answer for Ubuntu.

I propose the following addition:

  1. If the answer is untested and could work on ubuntu (such as being from a child-distro) and is clearly labeled as unverified on ubuntu that it be left standing until confirmed by the community to either work/not. My reasoning is that if it has been tested and confirmed to work on a related distro, I feel that solution at least has some potential to also be viable on Ubuntu more so than someone simply typing untested code into a reply and saying "try this". Furthermore, if the code does NOT work then OP has a known starting point to work backward from during comparison/debugging in order to help bridge the gap to a solution on Ubuntu.

I myself have had responses blocked/deleted that were said to be rules violations even though they were very well documented, using ubuntu repos, mostly distro-agnostic replies that happened to have one sentence stating that I had verified my code on a related distro and that it still needed confirmation under Ubuntu. Especially in cases where there is no valid solution to the asked question, blocking/closing replies for simply mentioning another distro without review the context seems a bit heavy handed. If we block these kinds of replies, the most likely thing that would happen is that people would either not mention what distro they tested it on or lie about it and say "Ubuntu" instead of "XYZ"; either way, it's worse than simply allowing mention of another distro as long as the focus remains squarely on Ubuntu.

| |
  • 1
    I'm sorry to hear you've had well-documented answers deleted because they mentioned not being tested on Ubuntu, especially considering how so many answers aren't tested at all. Even under the policy articulated there, it seems to me that what you're describing is not something that should have happened. Would you be willing to provide links to the questions where the answers you're talking about were deleted? Or, if you can't find them, are you willing to describe them, and/or okay with it if a moderator looks them up (if there's a mod willing to do so)? – Eliah Kagan Jul 9 at 1:24
  • @ Eliah Kagan, sure post was in response to this question but I assume only moderators can see it now. FWIW, I have tried to edit it to remove all references to the "unmentionables" but when I click the undelete link it says "Vote to undelete this post? (5 votes remaining)" followed by "A moderator has deleted this post and it cannot be undeleted" so that sounds like it is is probably not coming back. – zpangwin Jul 9 at 1:53
  • Also, to be completely fair / transparent, I was a little touchy about being flagged in this way and was not entirely civil when I disputed the rules (tone more so than language). I'll own that and admit I was in the wrong there but I still dispute that my post should have even been considered as a rules violation in the first place. I am of the mind that mentioning a thing does not constitute the conversation being about that thing. – zpangwin Jul 9 at 1:54
  • FYI, all users with >= 10k reputation can see deleted posts. One benefit of this is that such users can identify posts that were wrongly deleted :) – Zanna Jul 9 at 3:02
  • Hi everyone, that particular answer was downvoted not because something unmentionable was mentioned, but because it clearly stated that it has not been tested on Ubuntu. The OP, zpangwin, expressed confidence that the answer would also work on Ubuntu, but that confidence seemed groundless to me. Generally, if an answer says it's been tested on XYZ OSs + Ubuntu, I accept it, but I rarely see reasons to accept answers only tested on Zorin, Elementary, Popos, Debian, Mint, Kali, etc. I'd be glad to retract the downvote, and upvote the answer if and when the steps are tested on Ubuntu. – mikewhatever Jul 9 at 15:37
  • Forgive me if I am speaking out of place/am ignorant of things here but would it be possible to request new technical features that might allow some sort of middle ground? It occurs to me that deleting and downvoting mechanisms are really designed with other purposes in mind. And while perhaps they can be used in place of a missing feature that having something like a banner or tags appearing on the top of answers that say things like UNTESTED / NON-UBUNTU / VERIFIED / etc might be potentially a better option for everyone involved. Thoughts? – zpangwin Jul 9 at 19:16
  • ^ To clarify on the above, I mean to suggest the feature for both users and mods, not strictly one or the other. In the example of my previous post, I would have been happy to tag it correctly as something of a RFC (Request for Comment) / Unverified answer and in fact I did try to call it out as such. But perhaps tagging/banners could do very simply and with more visual attention than a user simply typing into a text box current fails to do. – zpangwin Jul 9 at 19:32
  • 1
    If people wish to downvote for such reasons, that is not unreasonable. Downvoting is the appropriate action in any case where you think there is some technical problem with the post. (However, requesting features is certainly welcome if you think they would make anything better - you can post feature requests here on meta) – Zanna Jul 10 at 2:09

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .