Change all occurences of uid with id

[bidoubiwa]

Since uid is the same in definition as id I was wondering if for simplicity we could not just change the names of the field to id.

I also, and this is only a feeling based on no evidence, that uid is used in a context of a simplified generated uuid. So I would think that uid would represent a sort of generated unique slug like 123abc45.

[dichotommy]

I agree that 1) uid is unnecessarily complicated in some cases 2) uid suggests a unique value generated by Meilisearch. I'm not sure that changing all uid to id is the best solution though.

As far as I am aware, indexes and tasks are the two objects in Meilisearch that have the uid property. Am I forgetting anything else? 🤔

What stands out to me is that index uid is user-set, whereas task uid is automatically generated by Meilisearch. I think it could benefit us to distinguish between these. For example, if we replaced index uid with name, I think that would immediately suggest a property set by the user.

(I believe there is already a name property attached to the index object but AFAIK it does nothing)

On the other hand, I think task uid could probably remain as is. Changing it to id would indeed be more simple. But from a documentation perspective, it could create confusion with the primary key, which is sometimes referred to as the document identifier or document id (and for inference, automatically looks for a field called id in documents).

WDYT @bidoubiwa ?

[dichotommy]

Edit: I forgot about dumps which also have uid. 😅

[bidoubiwa]

I like name for index! It does indeed imply that it is user set very well. As in SQL a name must be unique (even if in this case you can update it but that is not very important I think).

In the other cases, tasks and dumps, uid are indeed generated! Now the uid of the dumps is more in the sense of a uuid -> "uid": "20200929-114144097" where the uid of a task is more of an incremental id => "uid": 1, "uid": 2

[brunoocasali]

I like name for the index as well, our MongoDB friends use it too: https://www.mongodb.com/docs/manual/reference/method/db.createCollection/#mongodb-method-db.createCollection

But my feeling when I read name in a json object, I feel that I (as a user), can add anything here like:

{
    "name": "my beautiful index",
    "createdAt": "2022-04-14T19:59:29.226971255Z",
    "updatedAt": "2022-04-14T19:59:29.228628255Z",
    "primaryKey": "id"
}

{
    "uid": "my-beautiful-index",
    "createdAt": "2022-04-14T19:59:29.226971255Z",
    "updatedAt": "2022-04-14T19:59:29.228628255Z",
    "primaryKey": "id"
}

Without respecting any guidelines like these https://docs.meilisearch.com/learn/core_concepts/indexes.html#index-uid.

So, for me, the only change I would consider will be the tasks, summarizing:

• indexes - uid
• dumps - uid
• tasks - id

[bidoubiwa]

Yes I did not considered this "name": "my beautiful index",.
id, uid and name have their flaws. uid implies auto-generated complex number, id implies incremental and generated number and name let you think that you can name it with spaces (or other weirdness).

I looked at how mango tackle this issue, they created a naming restriction page that I found on your link. It explains how a name should be formatted.