You must log in or # to comment.
I just wrote something similar for decoding binary asm instructions.
If you have the time it’s a good solution!
Why don’t you just ask Chat-GPT o3 every time? Works like a charm!
Because there are better random generators
According to this code, “CEREAL” is a valid Roman numeral which equals 154. Great job!
Since Roman numerals have an upper bound, the time complexity is always O(1).
They forgot “CM” so this doesn’t work for any number that ends in 900s
No, M will be replaced by DD and then CD will be picked up, so it will go
- CM
- CDD
- CCCCD
- CCCCCCCCC
- …
Still linear time at least, could always be much MUCH worse
True. Lost opportunity to blow things up with useless recursivity
The word you’re looking for is recursion (see recursion).
Thanks. I knew something was off
Nah, I’d like to un-see recursion. It was way overblown on uni, I barely ever use it.
Recursion is amazing for a small selection of problems. Most of the time you don’t need, or want, it. When it is useful though, it tends to be really useful.
I don’t understand people’s issue with it. I always found it easy. Maybe that’s why I feel this way. Maybe if you find it challenging you want to avoid it, even when it’s a good solution.
Most devs I know like recursion. Trouble is that many popular languages don’t support tail recursion, but throw a stackoverflow error after a few thousand levels. So you have to keep track of max recursion depth manually, and it starts to look like a complicated solution
Most devs I know like recursion. Trouble is that many popular languages don’t support tail recursion, but throw a stackoverflow error after a few thousand levels. So you have to keep track of max recursion depth manually, and it starts to look like a complicated solution
You mean, like this?
I think, their point (and also my experience) is that you get taught about it in university a lot more than about simple loops, so it feels more important even though you rarely use it in reality.
Same thing goes for linked lists and inheritance…
Linked lists are encountered somewhat frequently in low level systems programming.
There could be a hidden quadratic cost because the string needs to be reallocated and copied multiple times.
Not quadratic in the length of the input. Assuming replace is linear this is also linear
Not if I don’t see it.
This is the spirit
unary!
IIV becomes IIIII, hm?
IIV would never be used. In Roman numerals at most one smaller unit can come in front of a larger one. The code doesn’t do any validation though.
While it doesn’t say anything about IIV specifically, they sure got creative enough to sometimes subtract more than one of the smaller units from a larger one.
Yes, that does demonstrate my point.
You missed “CM,” which was common in copyright statements in the 20th century.
No, they didn’t.
CM becomes CDD, which becomes CCCCD which becomes CCCCCCCCC.
It’s not too bad, it’s readable and easily optimised by adding intermediate sums and removing whatever power of 10 you’re working on.
My first thought was that it’d be a great oracle for randomized testing.
until(original=new) { run convertOriginal }
This isn’t sufficiently enterprisey for Java. There should be a Roman numeral factory followed by relevant fromString and toInteger methods.
Ugh. Literally refactored multiple factories into straightforward functions in the most recent sprint where I work.
Someone saw a public factory method which was a factory for a reason and just cargo culted multiple private methods using the same pattern.
Should do a regex find all then iterate over each chunk recursively until unchanged.
there was no regex in ancient rome
Actually there were seven kings prior to the establishment of the republic, at which point they expelled the rulers… a reg-ex if you will.
It also works the other way round: wanna convert Arabic n to Roman? Just write n times ‘I’ and revert these replacement in inverse order.
I don’t know what happens when the substring overlaps. Like for the number 6, will it replace the first 5 I’s with V and end up correctly with VI or the last ones and come to IV? I would guess the former and maybe you know but I never thought about it before
Also does not handle ‘IIIIIIIII’ -> ‘IX’ properly
If the substitution went right to left it might work.
It does.
I’ve written it that way and it works, as in it will replace left to right and you replace iiii to iv after iiiii to v
Makes sense but it will fail at 9 (VIV) it would only work for 9 if the replace went from right to left or the V and IV statements were exchanged but in both cases, 6 would fail
9 is IX though, and that works.
6 works fine, as it replaces the first set of 5 I with V and then there’s nothing to replace.
I’d written it in typescript for all it’s worth; go ahead and try it yourself :)
Does 9 really work? Wouldn’t it be:
IIIIIIIII -> VIIII -> VIVI like your questions about this and they all seem fair but I kinda wanna encourage you to go ahead and write it yourself; it’s a fun way to convert into Roman numerals that both is and isn’t intuitive at the same time.
No, cause you do the replacement from large to small, I.e. you’d first check for 10 I to replace with X (none found); then replace 9 with IX (check), then check for 5, 4 and so on.
I noticed my “and so on” is literally a noop here so yeah.
The original doesn’t have an extra check for 9 and it works for Roman->Indioarabic because it’s:
IX ->IVV ->IIIIV ->IIIIIIIIIBut the other way around, you need an extra step for 9. That’s where our misunderstanding comes from.
Depending on the language, you may be mutating the input value, which isn’t great.
I’m pretty sure it’s Java (due to the syntax and Eclipse editor default color scheme), so that isn’t an issue
Oh right, my Java is a bit rusty. But if it was Javascript, that would have been a problem.
