===== I put this last part of my answer on top because it might be what visitors are looking for. =====
Coming back in the “final end” with the “true solution”:
As so often I was badly informed (and though the only one to answer). As a side-effect of taking part in another forum I learned from “MrProgrammer” and from “atzi” that Calc (AOO and LibreO as well) since long was trimmed to be smart enough for the task. You simply have to implicitly describe the task by giving 2 consecutive cells and selecting them both before filling down by dragging the handle. Try it with.
XX-191-0100
XX-191-0101
getting next:
XX-191-0102
You may even define a different increment this way:
XX-191-0100
XX-191-0105
e.g. with an increment of 5 (or -5 as Calc will still interpret it):
XX-191-0110
There will, of course, come up other cases where the smartness of this feature won’t satisfy somebody. Please confer this forum thread for your further studies. I got there from this one.
Hope this is still of interest for you.
===== Next coming the originally first part of my answer. =====
Why?
The ‘serial’ filling is expected to also work with texts ending in a part that can be recognised as a decimal integer number. If we delimit this part in a “greedy” way, meaning in this case, to include the minus ("-") sign which is necessary for entering negative integers, the counting upward from -0110 will correctly produce -0109 as the next value. (Does MS not know that?)
You don’t like that? Well, it is correct, but there are workarounds if “less correct and more as I’m used to” is intended. For a detailed advice I might need some additional information. In specific: Are you allowed to replace the minus (hyphen) by another kind of a short-line-character, by a CHAR(173), e.g.? This will be treated as a non-numeric character. If this is to your disposition you may also use “#” or “~” or something else. Just not the numeric minus sign. You should realise that spreadsheet software comes from the engineering field and is still largely used there.
Please, don’t judge from a short visit. Your judgement as an experienced user of Excel might be biased by some MS specifics that aren’t simply correct at all. But: If you want a software acting exactly as MS Excel does you will need Excel and nothing else!
Alas! I remember complaints from Excel users over decades now that Excel itself sometimes doesn’t know how Excel works (or saves documents) after the version changed or someone made a decision in the dark.
====================================================
Coming back in the end (after some pursuit of happiness):
There is attached an example Calc document ask44711IncremeningProblem001.ods which is containing a small engine which may help in a few cases. It’s not satisfactory at all, I think. The document is also containing a few words of comment and a request for telling me how Excel is behaving in a specific situation.
I cannot solve the immanent dilemma. It is actually a case of expecting a specifice software behaviour seen as the one to expect from a specific point of view not having in mind that there may be other specific points of view expecting another specific behaviour. A concept to handle such cases should be to prefer a decision which might minimise conflicts with other preestablished concepts, emphasising logical ones. As far as I understand Calc chose this way. An actual tendency is to guess what most users might suit under some not expressly given limitations. I personally could well live with abandoning a “smart behaviour” in cases where smartness may not be indisputable. That’s not the way things go, however. More likely is, we might get a request for a feature to automatically increase/decrease the middle part of a text containing digits - or the front part. I don’t like to take part in the struggle for he correct endianness.