reduce (integers, mixed ordering) does not terminate

[ren]

Here is a suspicious example of reduce not terminating:

ring r = integer,(t,x(1..5)),ws(1,-1,-1,-1,-1,-1);
poly h = 3*t*x(2)*x(4)-21*t*x(1)*x(5)-12*t^2*x(1)*x(2)-2*t^2*x(1)*x(3)-6*t^2*x(1)*x(4);
ideal G = 
x(2)+6*x(4)+7*x(5)+4*t*x(2),
x(2)+8*x(3)+7*x(5)+t*x(2),
t*x(1)+3*t*x(2)+t^2*x(1),
t*x(4)+21*t*x(5)+12*t^2*x(2)+2*t^2*x(3)+7*t^2*x(4),
2-t;
G = std(G);
reduce(h,G);

What makes it suspicious is, if you are using a debug version and enable option "teach" (see attachment), you notice that it keeps reducing the same degree 2 monomials in x but with increasing degree in t, which keeps increasing because of 2-t.

This is not what I would expect. I'd expect some leading ideal to stabilize and the algorithm to terminate.

[ren]

output with "option(teach);"

[ren]

the bug seems to be already fixed on adi's version.

[ren]

still does not terminate, mistakenly checked an easier computation (which did not terminate in Singular but in Adi's version).

[hannes]

fixed with 67e0dc31cecb425433f640594911fc9df2b15076 tests: New/stdZtests.tst