That's my understanding as well

D is actually a Least Common Multiplier, not a common denominator.

In fact D (in Rust an in some other languages) is a "Least Common Denominator" hence a common denominator.

So I read the title and skimmed through the description ... and thought "okay, it's just reducing fractions" ... I read the example and it looked like the exact opposite

"Réduire des fractions au même dénominateur" is misleading: "reduced fractions" don't have necessary a "smaller" denominator.

D is quite large and I didn't notice it was the same ... reading the Discourse here I'm not the only one to find this description confusing

Problem is: before coding, it is better to carefully read in order to clearly understand the task...
Unfortunately it is a quality badly shared. Skimming through the description saves you a few minutes to lose a lot more later.

Thanks for your post.

The title is "Common Denominators". The word common shows clearly that it has to be the same for all elements!
Moreover you have below:

You have to produce a result in the form:

(N_1, D) ... (N_n, D) 
or
[ [N_1, D] ... [N_n, D] ] 
or
...

and everybody can see that D is the same for all elements.
You take Rust. You should have seen in "Sample Tests" that in each test D was the same for all elements.

@ThibaultLemaire: Now, it worked. Could you please check again.

then don't resolve the issue before you update the version.

You're right. But at the moment it is not possible to update because of an error of codewars...

Better to better your solution...

OK I know that since long but 1089 guys (out of 17631) have passed the C++ kata so nothing can be changed (limit for changing is 500 at CW).

If this weren't a 9 line, self-contained function maintainability might be more of a concern than efficiency.
But that would beg the question "Why isn't that its own function?"

I would certainly include comments with this code, but I stick by it!