If a user posts a code snippet/mwe stating a faulty behavior, but the code actually perform the correct action, is there something akin to a "Can not reproduce" type of flag? Example: Line/arrow using tikz

As stated in the comments the code already behaves as expected, but what happens to the thread - does it stay open or does someone close it?

  • 1
    Somebody comments. We wait to see if the OP agrees, disagrees, whatever. If the OP responds and says something like 'silly me!', it gets closed. Otherwise, the question either takes a new direction (based on response) or waits until nothing happens and it eventually gets closed.
    – cfr
    Commented Oct 12, 2015 at 3:35

2 Answers 2


In this case, the post shouldn't be closed first.

As @cfr explained in comment, the OP can have the problem and not knowing why this doesn't work for him/her.

Sometimes the error is a old TeX system and updating it to a current version is the best way to solve the problem.

Sometimes the MWE first provided by the OP didn't show the proper behavior: the OP made it quickly without compiling and you have to wait to see if a package use can be in conflict with the code.

So my answer is:

  1. Comment and say you can't reproduce the error
  2. If the OP modifies the code, retry and go to point 1 if needed
  3. If the OP admits a mistake on their side, vote to close, or flag the question for moderator intervention saying describing why.

@cfr pointed in the comment: the system can make an automatic deletion of a question: Enable automatic deletion of old, unanswered zero-score questions after a year? if the OP doesn't respond.

  • 2
    Might be worth adding that there is a system for ensuring such posts get closed eventually i.e. if the user never responds, they'll eventually get closed as unclear either when somebody randomly notices or in an 'answer the unanswered' sweep.
    – cfr
    Commented Oct 12, 2015 at 17:31
  • @cfr : Great info about non-response posts. However, how long is this "eventually" - use this post as an example tex.stackexchange.com/questions/185798/… As far as I can see, it's been non responsive for about 16 months, has no mwe and not responded to comments. Commented Oct 13, 2015 at 19:29
  • 3
    @FredrikJohansson 'Eventually' is like TeX's glue - infinitely stretchable. However, I think that question may possibly be best closed as a duplicate. It is reasonably clear and closing as a duplicate is more useful, perhaps.
    – cfr
    Commented Oct 13, 2015 at 21:03
  • @cfr +1 for "... like TeX's glue - infinitely stretchable". :-) So, in a case like this, who does the actual closing? The help sections are clear enough as to what type of questions get closed, except for unresponsive q's, and also not so clear on who closes questions in a case like this. I appreciate the time you've put into a better understanding of the topic. Commented Oct 13, 2015 at 21:15
  • 1
    @FredrikJohansson It is a bit complicated. A mod can close, of course. So can a user with a gold bad in a tag which was originally applied to the question. Otherwise, users can vote to close it. If one person does this, the question gets sent to a review queue and other people may vote there or if they just come across it. If enough people vote to close, it gets closed. It then stays closed unless it gets re-opened.
    – cfr
    Commented Oct 13, 2015 at 21:28
  • @FredrikJohansson I'll add that ones a month, we have an event: answer the unanswered, and everybody can participate. This consist to answer but also close if needed this kind of question ;-) Commented Oct 14, 2015 at 5:59
  • @RomainPicot :Thanks for the info. "answer the unanswered" sounds like a hoot... ;-) How is it marketed - that is, how do I get notified about those events? Commented Oct 14, 2015 at 11:50
  • 1
    @FredrikJohansson it's advertise on on TeX.SE in the right or in chat. Probably with other method that I've not in mind now too Commented Oct 14, 2015 at 15:10
  • @FredrikJohansson Stop by in our little chat room once in a while, you'll note the event :-) by the way, just by voting to close, a question doesn't get closed automatically with gluey time stretch. There is a limit of (i think) a few days, if not enough votes are recorded, the Q stays open.
    – Johannes_B
    Commented Oct 15, 2015 at 17:09
  • @Johannes_B so even though I may find an old question and vote to close it, there's little chance of it getting closed anyway? Unless someone else notice the same question during these few days? Commented Oct 15, 2015 at 17:38
  • 1
    @FredrikJohansson The chance of not closing is pretty low. As percusse might confirm, we have a few robo-reviewers on board. You could vote to close a very clear question with very good answers, and it probably would be closed anyway. Just for the sake of reviewing stuff. :-(
    – Johannes_B
    Commented Oct 15, 2015 at 17:46
  • Just noticed the Upcoming Event in six days notice in the right hand side. Wonder what the rules are... Commented Nov 2, 2015 at 8:11
  • 1
    @FredrikJohansson you can ask in the chat room and a mod will answer, or you can come and see :-) Commented Nov 2, 2015 at 9:17

In addition to the "loop" posted above by Romain Picot:

In case the unreproducible error stays visible on one side, but is not to detected on the other side, I strongly suggest to send this person to the Bugtracker of the used Software and/or Linuxdistribution (for example the Debian or Ubuntu bugtracker) and ask them to have a look at the discussion regarding the issue discussed at stackexchange.

In many cases unreproducible errors are "real"[¹], they just don't show under any conditions. It may happen for example that one problem occurs only in conjunction with other packages installed, or only if an underlying library has build in version x, or only if you are on a certain kind of hardware. Having a broader overview of the used components is only possible if people know this system from heart. Sometimes these people that can help better with system debugging than the "avarage tex wizzard" ;)

Thats why software projects in general and Linux Distributions in particular have Bugtrackers - please do use them.

¹: not-that-serious-sidenote: compare Heisenbug at Wikipedia 
"a heisenbug is a software bug that seems to disappear or alter
 its behavior when one attempts to study it" 
(Wikipedia contributors, "Heisenbug," Wikipedia, The Free Encyclopedia,
(accessed October 22, 2015).) 
  • 3
    While this is commonly true for general software, it's very rare for LaTeX, in the sense that the cause for the problem in LaTeX is usually a corrupted installation (such as a mixture of two versions, one package updated and another one not, only an editor being installed but not the distribution etc.). Also, when the code is not complete (which is what's being asked here mostly), it's very likely that something in the code that wasn't shown to us is the cause of the issue.
    – yo'
    Commented Oct 22, 2015 at 14:16
  • @yo': "a mixture of two versions, one package updated and another one not" and exactly this should not happen if using a distributions installation ;) Commented Oct 23, 2015 at 13:34
  • 2
    Oh you mean like the Debian's one which is always old?
    – yo'
    Commented Oct 23, 2015 at 13:43

You must log in to answer this question.

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