Skip to main content

Timeline for answer to What can we put in a question template to help people ask better questions? by user4639281

Current License: CC BY-SA 3.0

Post Revisions

39 events
when toggle format what by license comment
Jun 3, 2020 at 15:29 history edited CommunityBot
Commonmark migration
Nov 8, 2017 at 23:09 comment added jrh @TinyGiant huh. Well, I can't really ask for much more than that. It's even at +72/-1. Well, thanks for trying.
Nov 8, 2017 at 22:25 comment added user4639281 @jrh this is the quickest example I can link (its one of my highest voted answers) but it is all over meta.
Nov 8, 2017 at 18:46 comment added jrh I'd suggest making a meta post on this "code writing service" topic, I don't think I've seen a post that really gets directly to the point and discusses some of the downsides of not allowing those kind of posts (so far I've only seen "zero effort questions are bad"; maybe a bunch of other kinds of questions get lumped in as collateral damage).
Nov 8, 2017 at 16:14 comment added user4639281 @jrh Such questions have never been off topic. It is a shame the some users misunderstand the purpose of the system, and in turn they mislead others. Those users doing this actually generally believe that they are being helpful.
Nov 8, 2017 at 13:15 comment added jrh @MarkAmery Just out of curiosity, where did the resistance towards SO becoming a "code writing service" come from to begin with? I like your explanation about when "write some code for me" is okay, and I'm honestly surprised that it's an unpopular idea. Now that docs failed, does that mean that we're back to old blogs, SO posts from 2009, and old books for example code? Doesn't really sound like that fits with the original goals of the site.
Nov 6, 2017 at 21:57 comment added Peter Cordes @MarkAmery: The main thing that makes asm special is that everything tends to be less generic than in other languages. The exact details of e.g. problem size might allow different optimizations. So there's no generic best answer for comparing strings in x86 asm. And I/O facilities vary by platform (libc or not, Linux system calls, BIOS or DOS calls, etc.), so "print a multi-digit number" questions aren't all duplicates of each other. Except they really are because the algorithm for repeated division by 10 is the same... This is one kind of "specific detail" that gets problematic for dups.
Nov 6, 2017 at 21:43 comment added Mark Amery @PeterCordes ... there are some clearly bad kinds of homework/beginner-level questions. Outright duplicates are straightforward: close 'em and move on. The category I hate are the ones you describe as having "specific requirements that would make it not a duplicate". This often means, in my experience, that the asker is really asking two or three good (but duplicate) questions in one, but lacks the skill to pick them apart, Google each of them, and stitch the answers together. They're bad questions, but can't be closed without being "creative" in interpreting the close reasons.
Nov 6, 2017 at 21:38 comment added Mark Amery @PeterCordes With the caveat that I've never written assembly and might be missing assembly-specific nuances, I think I'd break the category of questions you're describing down a little. If the question is well-framed, broadly applicable, and not a duplicate, then I don't think the fact that it's "simple" should count against it - there's nothing innately wrong with questions being basic. Even if the question is directly answered in the language's docs, answering it is a great opportunity to introduce beginners to those docs with a link and a quote. On other other hand... (TBC)
Nov 6, 2017 at 21:31 comment added Peter Cordes @MarkAmery: Where do homework questions fall into that? In the assembly-language tag, people frequently post their homework assignments and the task is often simple and common, but I still feel like taking it at face value and answering wouldn't be great. You're right that the details are in the framing; often there are specific requirements that would make it not a duplicate of a basically similar question. It also feels wrong to write code for someone that can't write it themselves, rather than answering someone who could write it, but probably not well and would take them some time.
Nov 6, 2017 at 21:26 comment added Mark Amery @PeterCordes in other words, I consider the categories of "write some code for me" and "debug some code for me" to both contain a full range of great to terrible questions. The key factor, to me, is whether the question-framing effort has been put in to create a question that is simultaneously narrowly specified and broadly applicable. In the debugging case that generally means showing a common error using an MCVE, but it's equally possible in the "how-to" or "write me some code" case just by thoughtfully narrowing down the task.
Nov 6, 2017 at 21:21 comment added Mark Amery @PeterCordes "I don't want SO to turn into a code-writing service either" - some folks will disagree with me, but personally, I'm happy for us to be a code-writing service for any code that satisfies both of two conditions: 1) it's short enough to reasonably belong in a Stack Overflow answer rather than a library (though that still might be 100+ lines in rare cases), and 2) it's a generic yet well-specified task that many people will want to achieve and Google for a solution to. The bad code-writing questions are the "please completely implement all of my business logic" ones.
Nov 6, 2017 at 21:16 comment added Peter Cordes @MarkAmery: You make a good point. When you put it that way, I think I have made that "what have you tried" mistake sometimes. However, when the question is "how do I compare strings in assembly language" or some other trivial task, it's not really a good SO question either. Like the algorithm is so simple that the surrounding code is going to take up many more lines. I don't want SO to turn into a code-writing service either. OTOH I don't mind showing an efficient way when the OP sounds capable of writing something, not just a clueless homework question.
Nov 2, 2017 at 9:02 comment added user3956566 I honestly think this is the best solution. I hope they adopt this. It should remain the asking template for users with <=20 rep.
Oct 31, 2017 at 16:01 comment added Antoine Pelletier There could be another separate type of tags, you would get to choose between : debugging, theory learning, or code improvement. And as it would be... forced ? maybe just highly suggested ? then answerers will always know what to expect from OP !
Oct 24, 2017 at 20:10 comment added user4639281 @MarkAmery I agree
Oct 24, 2017 at 14:28 comment added Mark Amery @TinyGiant ah, right, yeah. That happens, and it's awful. The "what have you tried" crowd and their misguided mission to eliminate all questions other than "debug my wall of code" from Stack Overflow are a disease that the community hasn't yet managed to expunge.
Oct 24, 2017 at 14:27 comment added user4639281 @MarkAmery I believe boltclock was talking about cases of how to questions where people request an MCVE, which then turns the how to question into a debugging style question, hence making the question less widely applicable.
Oct 24, 2017 at 11:02 comment added Mark Amery @BoltClock "My main issue with MCVEs is that they tend to artificially narrow the scope of otherwise widely applicable questions." - eh? I'm on your side that a debugging-based question template would be a bad thing, but this particular claim seems confused to me. An MCVE generally broadens the audience of a debugging question by changing it from "why does my particular 1000-line application throw a FrobError" to "why does a FrobError get thrown in this simple circumstance that applies to many people?"
Oct 21, 2017 at 0:14 comment added Shog9 StaffMod Sounds like prompts that'd help establish a level of detail or skill required would be a valuable addition then, eh @jrh?
Oct 20, 2017 at 23:53 comment added jrh @Shog9 I don't ask anything usually, almost all "How Do I" questions I've ran into tended to be answerable as-is. I usually give an answer at the level of abstraction they asked and then wait for further questions to get an idea of what they understand. If it's a coding question I usually make a quick test program or snippet to demonstrate. I could even give an answer to a massive question like "How do I make a Quake clone" by talking in terms of general concepts instead of raw code; they wouldn't have Quake at the end of it, but they'd be a bit better off, hopefully.
Oct 20, 2017 at 23:41 comment added Shog9 StaffMod What are your standard questions when a co-worker asks you such a question, @jrh?
Oct 20, 2017 at 23:31 comment added jrh @Shog9 I gave it a try but I had a hard time coming up with any guidance for "How do I" questions that didn't seem too restrictive (I couldn't in good faith post a rule that would have prevented a post I honestly found useful). The only clear cut rule I could think of is "Don't post duplicates and be as specific as possible", but that doesn't really prevent most of the common things that should be avoided. So it sort of just came right back to "if you post something too time consuming, you probably won't get an answer.". Unfortunately that's all I've got, sorry.
Oct 20, 2017 at 5:08 comment added Passer By @jrh In my experience, there are a lot of conceptual questions would benefit from a MCVE. The reader understands that the example is an example and not the entirety of the question, but having an example illustrates clearer what words alone might not. Like this
Oct 19, 2017 at 17:51 history edited user4639281 CC BY-SA 3.0
added 3 characters in body
Oct 19, 2017 at 9:04 comment added Jean-François Corbett I took the liberty of emphasizing your main point. It was in the last line of your post, and I actually missed it completely while skimming. I was about to post an answer with the exact same suggestion, but decided to do a text search for "type" just to double-check that no one else had posted the same thing already. That's when I found yours. So, anyhow, +1
Oct 19, 2017 at 9:02 history edited Jean-François Corbett CC BY-SA 3.0
emphasized main point
Oct 19, 2017 at 5:26 comment added BoltClock Mod @Shog9 My main issue with MCVEs is that they tend to artificially narrow the scope of otherwise widely applicable questions. Yes, so do most askers anyway, but I don't fancy being restricted just because new askers (I specifically use that verbiage because I've seen 20k users who then ask a question for the first time and it's bad) aren't used to how the site works.
Oct 18, 2017 at 17:51 comment added Shog9 StaffMod Y'know, this doesn't have to focus on debugging questions... I kinda suspect MCVE showed up a lot in the mentoring stuff because it's an easy (and often-useful) bit of advice, whereas improving a "HOWTO" question kinda requires a bit more domain knowledge. If you can think of prompts that'd be useful for not-debugging questions, propose 'em! I did...
Oct 18, 2017 at 15:14 comment added jrh Also, for those of you that want to only accept questions with an MVCE on SO, is your theory that all of these conceptual sort of SO questions would fit better on Softare Engineering? Are you thinking that we should migrate non-MVCE (e.g., "How Do I" questions, or "help me understand this concept") questions to SE.SE? Would that even work? I'm not really sure it would work, to be honest. Or are you just looking for mass deletion/closure of "How do I" / "I have a question about a concept" questions? I'm just trying to figure out what the future is of those sort of questions.
Oct 18, 2017 at 13:54 comment added Don't Panic Yes, it would be great if the templates would help people to ask actual questions, not submit help tickets. There really aren't that many categories of on-topic questions, but a good template for each category of question would be counterproductive when applied to a question of a different category.
Oct 18, 2017 at 12:22 comment added jrh @BoltClock Yeah, I agree, it's a drag. IMO "debug my code for me" questions are a very weak subset of what SO used to be able to accomplish. There's some useful information in that genre but the more conceptual questions tend to get more done with less; even a straightforward "How do I..." question is probably more useful to readers than 8 paragraphs of boilerplate "why your code doesn't work". It's also a lot more awkward to write self answered QA if SO only accepts "debug my code" questions.
Oct 18, 2017 at 8:09 comment added Martin James @BoltClock indeed. It's all a conspiracy between Hollywood and the profs/TA's to portray software development as 'writing code that works immediately if you are a good programmer', rather than engineering where continual loops round testing/debugging are always required to get software fit for purpose:( This is blatantly obvious from the many posts: 'I dont' understand - my code compiles with no errors or warnings, but still segfaults'. Myself, I would start by exterminating the current crop of profs/TA's;)
Oct 18, 2017 at 6:58 comment added BoltClock Mod This is what I'm worried about: Stack Overflow being reduced to "debug-my-code-for-me". Not the first time I've expressed this concern either.
Oct 18, 2017 at 2:43 comment added Makyen Mod @JonEricson, Perhaps what's appropriate, prior to showing a template, is to have the user select from a set of radio buttons as to why they are asking a question. For example: "My code doesn't work the way I want"; "There's a concept I have a question about";[some other options]; "My question is not specifically about programming" (direct them to some other site); and "My question is specifically about programming, but doesn't fit any of the above". Obviously, those are just off the top of my head, but it allows you to tailor the template to the type of question the OP thinks they have.
Oct 18, 2017 at 2:29 history edited user4639281 CC BY-SA 3.0
deleted 72 characters in body
Oct 18, 2017 at 2:21 history edited user4639281 CC BY-SA 3.0
deleted 72 characters in body
Oct 17, 2017 at 20:28 comment added Jon Ericson StaffMod In the mentoring experiment, we did see that an awful lot of questions from new users were related to code not working. It might be that simple templates will cause more problems then they will solve. That's fine, as long as we learn a thing or two. Whether templates work or not, we'll have a better idea on how to create a guided asking experience that really works. Figuring out what sort of question a user is asking might be the first step there. We'll see.
Oct 17, 2017 at 19:46 history answered user4639281 CC BY-SA 3.0
toggle format