Weird RecurrenceTable bug

Question

Bug introduced in 10.0 or earlier and persisting through 12.2 or later

The output of RecurrenceTable seems to lose a level when there are more than 250 points and one variable. Easiest to explain with an example:

Good:

RecurrenceTable[{n[t + 1] == 2 (1 - n[t]) n[t], n[0] == 0.1}, {n}, {t, 0, 249}]

(* {{0.1}, {0.18}, {0.2952}, {0.416114}, {0.485926}, {0.499604}, {0.5}, {0.5}, ... {0.5}} *)

Bad:

RecurrenceTable[{n[t + 1] == 2 (1 - n[t]) n[t], n[0] == 0.1}, {n}, {t, 0, 250}]

(* {{0.1}, 0.18, 0.2952, 0.416114, 0.485926, 0.499604, 0.5, 0.5, ... 0.5} *)

Having more than one variable works as expected:

RecurrenceTable[{n[t + 1] == 2 (1 - n[t]) n[t], m[t + 1] == 2 (1 - m[t]) m[t], n[0] == 0.1, m[0] == 0.1}, {n, m}, {t, 0, 250}]

(* {{0.1, 0.1}, {0.18, 0.18}, {0.2952, 0.2952}, {0.416114, 0.416114}, {0.485926, 0.485926}, {0.499604, 0.499604}, {0.5, 0.5}, {0.5, 0.5}, ... {0.5, 0.5}} *)

I've found a couple easy workarounds for this minimal example (e.g. use n instead of {n}, use NestList instead of RecurrenceTable), but I've got RecurrenceTable embedded in a general function that requires both a list of dependent variables and doesn't play well with NestList. I just wanted to see if others have this same issue, which I'll report to WRI. I'm using Mathematica 11.0.1 on MacOS 10.12.

Answer 1

@ilian's fix in comments above works fine: add Method -> {Compiled -> False}. I've also reported to WRI.

Edit: But you know what, compiling is kind of a nice feature, so another workaround is to post-process the results, like

Replace[
  RecurrenceTable[{n[t + 1] == 2 (1 - n[t]) n[t], n[0] == 0.1}, {n}, {t, 0, 250}],
  n_?NumericQ -> {n}, 1]