I am seeing the following issue on Windows 7 x64, with the new Mma 11:

I define the simple function below, and evaluate it. The first time is fine, but after that the kernel seems to have forgotten that I defined f:

enter image description here

Next, I simply re-evaluate the function definition and the next four lines, and I see this:

enter image description here

which is as it should be. Note that for each of the above, I proceeded by pressing Shift-Enter on the five Input lines, one after the other. If I save the notebook and simply choose "Evaluate Notebook" from the menu, things look fine. However, if I then manually ask for an evaluation of f in a new Input line and press Shift-Enter, I again get an unevaluated f back.

By now I have spent way too much time on this nonsense, but the bottom line seems to be that I need to evaluate my function definitions twice for them to "stick". Clearly, that's not the way this should work...

What is going on here? Can anyone confirm this behavior?

  • $\begingroup$ I cannot reproduce this problem on a Mac running Mma v11.0 $\endgroup$
    – Bob Hanlon
    Commented Aug 16, 2016 at 0:32
  • $\begingroup$ This function you defined works and no problem on my Win10 64 with Mma v11. $\endgroup$
    – Orders
    Commented Aug 16, 2016 at 0:35
  • 8
    $\begingroup$ (75284)? $\endgroup$
    – Mr.Wizard
    Commented Aug 16, 2016 at 0:53
  • $\begingroup$ @Mr.Wizard: You rock! That was it. $\endgroup$
    – Pirx
    Commented Aug 16, 2016 at 1:32
  • $\begingroup$ I cannot reproduce this on Windows 10x64 and Mathematica 11. The Suggestions Bar is enabled (gave problems in earlier versions!) and the $HistoryLength is set to 10. $\endgroup$
    – user36273
    Commented Aug 16, 2016 at 10:09

1 Answer 1


Solution (kind of...):

Mr.Wizard pointed to the answer: Disabling the suggestion bar solved the issue. To quote from the post he linked to:

The predictive interface (Suggestions Bar) is the source of many bugs reported on this site and surely many more that have yet to be reported. I strongly suggest that all new users turn off the Suggestions Bar to avoid unexpected problems such as massive memory usage, peculiar evaluation leaks, and broken assignments.

  • 5
    $\begingroup$ I feel sorry for the developers who wrote that thing as I bet they were given half the time they needed to do it properly. $\endgroup$
    – Mr.Wizard
    Commented Aug 16, 2016 at 1:44
  • 3
    $\begingroup$ There's definitely some very strange things going on with that bar. I see this as a very general problem with Wolfram products: Excellent, even uniquely powerful ideas, but execution is lacking, with poor "fit-and-finish", and limited attention to detail. Reported and confirmed bugs lingering in their software for years are but one of the symptoms. Oh well, nothing in this world is perfect. $\endgroup$
    – Pirx
    Commented Aug 16, 2016 at 2:24
  • 10
    $\begingroup$ Sadly I am becoming pessimistic about the future of the software; several releases have now focused on greatly expanding the standard library of functions while often ignoring the function and performance of existing ones or even the utility of the new ones, e.g. (55294). I feel that the software is in great need of a separation between the ever-expanding function library and the core language itself, the latter being long overdue for modernization and optimization, IMHO. (continued) $\endgroup$
    – Mr.Wizard
    Commented Aug 16, 2016 at 2:48
  • 11
    $\begingroup$ The core language/kernel would ideally be free or low cost, and the giant (and admittedly impressive) function library would be the major product. I think this would ultimately benefit the entire Mathematica ecosystem including of course WRI. $\endgroup$
    – Mr.Wizard
    Commented Aug 16, 2016 at 2:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.