Why doesn't datetime error out when an impossible result is calculated in R2024b?

For example, I'm running:

>> datetime ("03-Feb-2025","Format","DD-MMM-uuuu HH:mm:ss.sss")Warning: The format 'DD-MMM-uuuu HH:mm:ss.sss' contains fields for month (M) and for day of year (D). This will cause unexpected results when converting from text.See the datetime.Format property for a complete description of the identifiers used in datetime formats.> In verifyFormat (line 34)In datetime (line 188) ans = datetime  34-Feb-2025 00:00:00.000 >>
Matlab only issues a WARNING (not an ERROR) and produces an impossible result, i.e. 34-Feb-2025.
Is this expected?
(Setting « dd » instead of « DD” gives the correct result.)

Answer 1

[MathWorks Support Team]

Yes, this is the expected behavior.

The timestamp is parsed correctly and the underlying data in the datetime is just fine. You're just displaying it incorrectly, which is why this is just a warning.

>> t = datetime ("03-Feb-2025","Format","DD-MMM-uuuu HH:mm:ss.sss") % confusing format string>> t1 = datetime(2025,2,3); % avoid text parsing altogether>> t-t1ans =   duration   00:00:00

The Format property does not impact timestamp parsing - that's done using the InputFormat property. The Format just impacts the display. Change the Format to something sensible, and you're good to go.

>> t.Format = "uuuu-MM-dd"t =   datetime   2025-02-03