Looking for Cleanest Way to Check Indices

[paul_cardinale]

I am processing some data in which one of the data items is an index. It should be that each index is present exactly once.  I can assemble the indices into an array, sort it, then check that the element values match their indices.  But I also want 2 lists: One of indices that are missing, and another of indices that occur more than once.

Here is an example of a valid set of indices:

Here is an example of an invalid set of indices:

What's a nice clean way of getting the lists of invalid items?

[crossrulz]

Here's what I quickly came up with.  I used a Set to avoid duplicating replicated indecies and then a FOR loop that I expect to usually run 0 times to find the missing indicies.

[GerdW]

Hi Paul,

please define "missing indices"… (What's the valid range of indices?)

How large will those arrays become? Maybe using Sets/Maps/VariantAttributes might work?

[paul_cardinale]

---

@crossrulz wrote:

Here's what I quickly came up with.  I used a Set to avoid duplicating replicated indecies and then a FOR loop that I expect to usually run 0 times to find the missing indicies.

---

Thank you; I like that. I started implementing it, then I realized I should have given the bigger picture instead of just a piece.  I'm sure there's a better way to implement the whole thing.

I have a set of sockets that are identified by sequential zero-based numbers (i.e. socket indices).  Each socket is a member of a group.  Group IDs need not be sequential.
On my input, I have an array of clusters; each cluster containing a group identifier, and an array of IDs of member sockets.
On my output, I want an array with each element corresponding to a socket (array index = socket ID) whose element value is the ID of owning group.

I also want to generate a warning when the data make no sense, and to "repair" the data (of course it will be up to the user to read the warning and check the repaired data).

[altenbach]

Quick attempt at the original problem:

(note that "duplicate" simply means "more than one". we don't need to find or count all instances)

[paul_cardinale]

Thanks to everyone.  Here's what I ended up with:

[paul_cardinale]

This is a little cleaner:

[altenbach]

It is not clear how you do conflict resolution, but I guess the "Groups by Socket" is meaningless if there are still duplicates. For example it is not clear why you decide that the first element is 7 and not 2.

Here's my quick attempt. Same result (except first element is 2 instead of 7. Maybe less code, or at least easier to read). No guarantees. 🙂

[paul_cardinale]

---

@altenbach wrote:

It is not clear how you do conflict resolution, but I guess the "Groups by Socket" is meaningless if there are still duplicates. For example it is not clear why you decide that the first element is 7 and not 2.

Here's my quick attempt. Same result (except first element is 2 instead of 7. Maybe less code, or at least easier to read). No guarantees. 🙂

---

It doesn't really how conflicts are resolved; it's not possible for the S/W to determine what the data should be.  I just need to have a valid data set to send to the GUI and a message to the user so he/she will look at it and make corrections.

[altenbach]

... and a quick attempt to consolidate all that duplicate code for the error prep (just for fun 😉 ).