2

Is there any mechanism to check solutions again?

Look at this thread I cannot determine the error between hyperref and hypdvips

Problem was posted, problem was solved. But I am not able to compile this example, I have posted a very similar one, and also failed.

So, is there any mechanism of example which are solved and compiled for each update. In my opinion each solved example must run after each update of the system, isn't it?

Is this a good idea to ensure that solutions stay solutions?

| |
  • 2
    In one sentence: this is a nice theory, but it's impossible. – yo' Feb 17 '16 at 13:23
  • Why do you think thats impossible? – Peter Ebelsberger Feb 17 '16 at 13:26
  • Who would compile the examples? Who would mark which code should be treated this way? Who would include the information about how to compile the document? How do you verify that the example produces the desired output; a lot of code does not fail with an error, but still does not produce the desired output. Test suites and in general testing is a big science and costs a lot lot lot of time. – yo' Feb 17 '16 at 13:35
  • I think, that it is worth to think about it, even if you don't reach 100%. we always demand a MWE, so there is a compileable code. If you can provide a solution, you can save this example onto a server and let each example compile on every update. Checking if the state is identically to the previous resultfile should give you the information about changed results. But thats just a Idea in 2 Minutes. – Peter Ebelsberger Feb 17 '16 at 16:00
10

No.

The best way to bring attention to this is to place a comment in the answer that you've deemed to "not work anymore" and request an update from the original poster. If that doesn't work, update the post yourself. If a new solution presents itself using some completely different method, post a new answer so it can form part of the existing knowledge base.

One great example of this might be questions around - the class experienced a major change in design when the code was "refactored into heads, body, foot and icon substyles" for version 2 (see the change log, 28 July 2015). It is likely that many of the patches, additions or manipulations suggested up to that point for versions < 2 no longer work as expected.

It would be impossible to suggest all users keep all their answers (and questions!) up-to-date to function as expected with the most recent distribution update, since package authors can change functionality at any time.

| |
  • Maybe you can see, that the suggested approach increase codequality and compatibility for the whole community. Your post adresss only the package authors, but they are most likely not aware that this work for some members increase maintenance of the users because he has forgotten that his package is widley used. In my opinion is any example a test of packagecombinations and or functionality. It is external testing of code. If the community is willing to think about this....in my opinion all codes will have more quality as a result of massiv testing and the feedback loop will be much quicker... – Peter Ebelsberger Feb 17 '16 at 21:22
  • @PeterEbelsberger Who pays for the massiv testing? – Johannes_B Feb 20 '16 at 7:51
  • Old versions of moderncv are still distributed, the bunch of questions for <2 is still useful for many. On the other hand, a package author can decide that a package/classe is incompatible with certain other packages. – Johannes_B Feb 20 '16 at 7:51
  • @Johannes_B who pays for providing solutions to this site? – Peter Ebelsberger Feb 20 '16 at 20:40
  • @PeterEbelsberger Nobody, meaning ... if one needs more attention, there is less time to do the other. – Johannes_B Feb 20 '16 at 23:37
  • Does nobody see the benefit of improving code and testing, really no one? What a pitty. I did not expect this response on this idea. It seems to me, that there is no willing to think about how to provide such a solution, there seems more energie to denial this idea. But its ok for me. Lesson learned. I will focus on reporting bugs as it seems to me that this is more successful to me. – Peter Ebelsberger Feb 21 '16 at 0:11
  • @PeterEbelsberger: "Unfortunately" I have 170+ answers for every one of your answers. Perhaps it would be feasible for you to peruse your ~30 answers every now-and-again to make sure they're up to date. Most others don't and solely rely on visitors of your questions/answers for usage and quality checks. If something doesn't work anymore, I would hope that the visitor would take the courtesy to comment and request an update. We all want to do our best and try that. It would be great if you could identify problems in existing answers and comment to the poster for an update. – Werner Feb 21 '16 at 5:29
  • No Werner, not you. I know you are a very active member and I writes this comment only reading the first 4 Words of your comment and reading your name! – Peter Ebelsberger Feb 21 '16 at 11:13
  • @werner my experience is over and over again, that exectuing your suggestion is anoying everyone. I have began to make a quality system locally and automatic check if desired output is working or none working.As jo' pointed out,I will loose all documents who compile fine but do not deliver desired output.I'm aware about this fact.But its much better to catch some erros than catch no one. And once again, I'm very aware that you answer very lot of questions here. It was only the generell disappointment that no one is try to find a solution that could be managable, but find reaseons why it fail. – Peter Ebelsberger Feb 21 '16 at 11:19
  • @PeterEbelsberger Package authors test their packages and the documented features. Some stuff on this site is a mere hack, or patch. It is pretty obvious to me, that patches are just patches and no solid solutions. – Johannes_B Feb 22 '16 at 10:31
  • On the other hand, people asking a question here have alrready started. Some of them with the wikibook, which is in need of much more fixing: How can we, as a community, improve the LaTeX WikiBook? – Johannes_B Feb 22 '16 at 10:32
  • @johannes_B One last question and then I will accept this answer, I am not satisfied with it but I don't want to anoy the community any more. I dont see that I am willing to do nothing and I will start to try to improve the code even if I am alone. Can you see nothing that would improve the code at this suggestion? – Peter Ebelsberger Feb 22 '16 at 22:54
  • @PeterEbelsberger I don't think anybody is annoyed. – Johannes_B Feb 23 '16 at 6:53
  • @PeterEbelsberger There is a question on the big meta that deals with the same issue: Answers get obsolete. When you see something that can be fixed, fix it. If a package introduces something new, somebody famliar with that package changes the two or three answers on site that needs changes. But you cannot rely on that and certainly can't do it systematically. I really do think, that the source (bad/wrong documentation for starters) needs fixing more urgently. – Johannes_B Feb 23 '16 at 6:56

You must log in to answer this question.

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