The purpose of this thread is to propose candidate topics for writing canonical questions about (or using old questions for this purpose), and discussing them.


@JM suggested creating a set of canonical answers that we can mark recurring questions as duplicates of. Please read about the details here.

I think it's time we start collecting candidate posts to be repurposed as a canonical questions/answers.

If you notice a recurring question, please post a new answer where you list all related duplicate posts from the site (that are essentially about the same problem, even if they're not marked as duplicates). You can list posts from SO and MathGroup as well---these will serve as evidence that the problem/question really belongs in a FAQ. You can provide any additional relevant details or suggestions about what the canonical version of the question should look like.

Then the community can review these topics, select the truly FAQ-worthy ones, and move to the next step of writing up a canonical version of the question and answering it.

share
1  
One of the things not yet clear to me in this context is what should be the granularity of these canonical answers. In other words, do we want a one-to-one or one-to-many relationship for cononical answers vs FAQ answers. I am in favor of the latter, since it often happens that similar topics are inter-related, and having a single large canonical answer with subsections can eliminate the need for cross-links etc. For example, we may have a canonical answer dealing with memory issues, and your example answer would be one subsection there. So, my opinion is that we will be better off with ... –  Leonid Shifrin Jun 1 '12 at 11:39
1  
... "coarse-grained" and comprehensive, but well-structured canonical answers, and then a separate FAQ which would mostly map more specific questions to (sub)sections of those canonical answers. This will also be beneficial for the readers, since they will get their answers in a proper wider context and learn more. This is actually nothing new, many help systems work in this way: special question -> generalize -> specialize again, but give a proper general context. –  Leonid Shifrin Jun 1 '12 at 11:41
1  
I wonder, is there a way to filter for all previously closed threads that were judged as duplicates? This would help finding recurring cases a lot. –  István Zachar Jun 2 '12 at 11:10
    
@Leonid Sorry about the late reply. I don't have any strong opinion on the matter. I think there's only one important thing to keep in mind: the purpose of these "canonical questions" is to point to them when closing something as 'exact duplicate'. So it's important to make sure that the questions we close can indeed be considered as duplicates of these. It would also be good if it were not too difficult to find them using search. I think if they're extremely general, that could be detrimental to fulfilling their purpose as general duplicates, but a little generality will only help. –  Szabolcs Jun 4 '12 at 8:05
    
@Szabolcs Yes, that's a good point. We just need IMO to find a solution which would strike the right balance, to avoid duplication of efforts on our side. It may be that we may be better off with the two-level structure: large "meta-canonical" answers, really comprehensive and explaining stuff in a wider context, and then very short canonical answers which refer to those for details and basically just list the solution, without much explanation. In this way, one wel-written "meta-answer" can serve a lot of "canonical" answers, and we won't have to endlessly explain basically the same things. –  Leonid Shifrin Jun 4 '12 at 9:04
    
Too lazy now to search for references, but a note to self or others that may want to compile Q lists: 1) Graph[] functionality not available or difficult to find out 2) Customizing Plot[] color schemes 3) Custom Tickmarks in Plots 4) Overcoming Limitations of BarChart[] 5) Evaluate[] usage 6) Formatting Numbers 7)Formatting cells 8) Producing animated output (GIF/AVI) 9)Limitations in sound treatment 10)Module vs With vs Block vs .. 11) Dynamic[] common issues 12) Control[] placement and usage –  belisarius Jun 10 '12 at 21:17
add comment

7 Answers

Running out of memory because of failing to set $HistoryLength=0

I think it's worth having a canonical question about this topic. It's probably worth making it even a bit more general and discussing way to conserve memory (starting with the simplest things such as $HistoryLength=0, up to advanced techniques like file-backed variables). I'd appreciate some comments on how general you think the answer should be (i.e. $HistoryLength only, or general advice about saving memory).

Questions where the solution was $HistoryLength = 0:

share
add comment

Plot multiple datasets/functions with two y axes

Judging by the recurrence of this question on MathGroup, SE and perhaps SO as well, I assume there is still great interest in such constructs, and people will ask this question from time to time as the original answers fade away.

share
add comment

There is a related family of questions that probably need a master, canonical answer addressing all variations in one place:

share
    
Perhaps this covers the same (or sufficiently similar to the) topic I've proposed below? Should these be merged? Or do you see a strong difference between the two approaches? I'm not sure about it, but anyway, you're the expert in tag evaluation :) –  István Zachar Mar 22 '13 at 17:47
add comment

You must log in to answer this question.

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