6

This is now bugging me too much to ignore it anymore. When I interact with a notebook that has dynamic content, for example moving a slider, Mathematica regularly brings other open notebook windows to the front. This means that while I am interacting with the given notebook, some other notebook (already opened, but in the background) pops up in the front, either blocking my view or just staying in the lower-left corner (if it was minimized before). The focus is not changed, so I can still interact with the content of the original notebook now in the back. When I continue manipulation, the front notebook disappears, just to reappear again a few seconds later. This is extremely annoying and happens "if the Manipulate [or other dynamic content] contains something that's computationally expensive" (Sjoerd's comment below, and I can confirm this), mostly when complex 2D or 3D plots are manipulated directly (e.g. 3D rotate with mouse) or indirectly (mouse driven slider changes parameters of dynamic plot).

For example rotating the following 3D figure with the mouse causes the behaviour to present itself. 

Plot3D[Sin[x + y^2], {x, -3, 3}, {y, -2, 2}]

Reproducibility is occasional, sometimes it happens fairly frequently, sometimes not at all. A more or less solid way to reproduce the error:

  • open a fresh Mathematica
  • open a new notebook (by e.g. ctrl+N; this will be the one that will pop to the front; this can be any notebook window)
  • open another notebook, paste above code
  • evaluate code (dragging the 3D plot now won't trigger the glitch, or only very rarely)
  • bring in front any other application, e.g. Firefox (though the glitch could happen even if no other application is running)
  • get back to Mathematica (by e.g. minimizing Firefox)
  • dragging the 3D plot around for some time (~5-10 sec) has a high chance to trigger the glitch, which can manifest as short "jumps" of open Mathematica windows where one gets in the front for a fraction of a second)

Does anyone else experience this annoyance?

What causes this behaviour?

This is still reproducible under Mathematica 9.0.1, Windows 7.

István Zachar
  • 47,032
  • 20
  • 143
  • 291
  • Is there anyway to get the code for the dynamic slider, so we can try to reproduce what you're seeing? – tkott Apr 10 '12 at 16:15
  • Is there any observable pattern? For example, the window shown in front is an "Untitled-" notebook; is this always the case? Does this happen as often shortly after Mathematica is opened, or mostly after it has been running for some while? Is there any situation in which the problem is always reproducible? – Mr.Wizard Apr 10 '12 at 16:32
  • @tkott, Mr.Wizard: It is not caused by a particular piece of code, but I gave an example that does cause it sometimes. The other window can be anything, even the Messages window I think, if it was checked once before. Mostly it happens after some time, but I managed to do it now with a fresh Mathematica. No, sadly there is no situation where it is always present. – István Zachar Apr 10 '12 at 17:03
  • I can neither produce this error nor recall it happening on v7, therefore I have tagged this question version-8. If anyone using a different version, perhaps on a different OS, also experiences this error, please remove the [tag:version-8] tag. – Mr.Wizard Apr 10 '12 at 17:40
  • @Mr.Wizard, you must be right, because nothing like this happened before 8. – István Zachar Apr 10 '12 at 17:51
  • I tried to reproduce on win/32 MMA 8.0.0, but to no avail. Perhaps I'm not trying hard enough to crash MMA :( – tkott Apr 10 '12 at 19:32
  • I have this quite often if the Manipulate contains something that's computationally expensive and have mentioned this before in passing either here or on SO. – Sjoerd C. de Vries Apr 10 '12 at 19:54
  • 2
    If I try your simple Plot3D code and keep dragging the plotted function quickly in circles invariably, within 15 seconds or so of continuous movement, one of the open windows pops up in front. Very reproducible. – Sjoerd C. de Vries Apr 10 '12 at 20:10
  • I cannot reproduce this on Mathematica 8.0.0.0 on Mac OS… – F'x Apr 10 '12 at 23:17
  • No issues here on MMA 8.04 Mac OS-X 10.74 – Nothingtoseehere Apr 10 '12 at 23:57
  • I cannot reproduce this with 8.0.0.0 on Linux. – celtschk Apr 11 '12 at 15:45
  • Well, I sense that this question will be open for a long, long time... – István Zachar Apr 14 '12 at 23:38
  • @IstvánZachar We can close it for as off topic you if you like :P Joking aside, shouldn't this be addressed to Wolfram support as well? I have noticed similar behaviour as well (mostly when the formatting notebook contents window shows up). – Ajasja Jan 09 '14 at 18:43
  • @Ajasja I did not send it to TS, as I was afraid that this is indeed to localized, version/platform dependent, ill-reproducible (?), marginal, etc. I am still waiting for a reasonable amount of backing from the community to bother with an official bug-report. – István Zachar Jan 09 '14 at 18:49
  • I cannot reproduce it Mma 9.0.1, Mac OSX 10.8.5 – Michael E2 Jan 09 '14 at 19:39
  • @Sjoerd Are you on v10? I am, and I can pretty reliably say that this annoyance is gone. Could you still reproduce it? If not, would you please post a short answer that it got "solved" (whatever the cause was), so that I can accept it? – István Zachar Jan 06 '15 at 09:48
  • @IstvánZachar I'm on V10 now, but I've also got a new laptop with a new OS (8.1 instead of 7), more memory, faster processor and graphics card. I don't see the glitch, neither in 10 nor in 9, but that does not mean that v10 on my old laptop would have worked. – Sjoerd C. de Vries Jan 06 '15 at 20:53
  • I may, perhaps, be not as experienced as some MMA users, however I believe there is an issue with the memory allocation and where these variables are getting stored on your machine. – e.doroskevic Oct 31 '15 at 01:44
  • @E.Doroskevic could you be a bit more specific please? Your comment is rather vague. – István Zachar Oct 31 '15 at 11:48

1 Answers1

0

Upgrading to a newer computer solved the problem. Perhaps it was RAM- or grpahics-card-related, we'll never know.

István Zachar
  • 47,032
  • 20
  • 143
  • 291