Other ordinals than First and Last

[Ezibenroc]

Cf. ProjetPP/PPP-QuestionParsing-Grammatical#95 (comment)

I don't see any way to describe a question such as "Who is the fourth president of the United States?" with the current datamodel.

We should add an operator for this.

Two alternatives:

• Nth(l,n) returns the n-th element of the list l.
• Minus(l1,l2) returns the list containing the elements of l1 which are not in l2.

We have Nth(l,4) = First(Minus(Minus(Minus(l,First(l)),First(l)),First(l))).

[yhamoudi]

We have Nth(l,4) = First(Minus(Minus(Minus(l,First(l)),First(l)),First(l))).

No way! (for Minus)

[progval]

+1 for Nth

[Ezibenroc]

Yes I find it very ugly, and this might trigger huge communications if one user ask for the billionth item.
But the operator Minus is more general and could be usefull in other cases.

[Tpt]

My 2 cents:

• "But the operator Minus is more general and could be usefull in other cases." Yes, but in our use case it may create very long trees (example: "Who is the 29th president of the United States?"). Do we want such trees? So, I think we may pay the cost of having two nodes (Minus and Nth).
• I don't like the name "Nth" (that is imho a little bit not clear). But I would like to avoid "at position" (too long). Does someone have a better name?

[Ezibenroc]

I don't like the name "Nth"

This is the name used in OCaml and Erlang!

[Tpt]

I'm not sure that these two languages have the best readability. It's not because some other people do something that this thing is good.

[marc-chevalier]

I think the two operators are useful, but not now. Wa can add the minus
later if needed...

And I like the name nth, it is much more popular in english than the two
examples can make think and, indeed, it is very clear.

[yhamoudi]

This issue was still opened here: #35

Here's what it says (i closed the issue):

position(l, i) should returns the ith element of l.
First(l) would became an alias for position(l, 1) et last(l) one of position(l, -1).

[Tpt]

I think we all agree on the use cases of the operator.

Two questions remain:

1. Position(l, i) or Nth(l, i)?
2. Should we tag First and Last as deprecated in order to remove them in the future or should we keep them indefinitely as aliases of position(, 1) and position(, -1)? (in both cases, datamodel libs may do some normalization in order to allow modules to don't have to deal with the two writing of the same thing).

[progval]

1. Nth. Position is ambiguous (it could be the index of an item of the
   list).
   I would also prefer Nth(i, l) because the list could possibly be very
   long and it would be less readable to have the index after it.
2. Yes, I think we should make them deprecated.

On 16/02/2015 15:05, Thomas Tanon wrote:

I think we all agree on the use cases of the operators.

Two questions remain:

1. Position(l, i) or Nth(l, i)?
2. Should we tag First and Last as deprecated in order to remove them in the future or should we keep them indefinitely as aliases of position(, 1) and position(, -1)? (in both cases, datamodel libs may do some normalization in order to allow modules to don't have to deal with the two writing of the same thing).

---

Reply to this email directly or view it on GitHub:
#51 (comment)

[Ezibenroc]

@progval: +1