Why use pagination tokens?

Question

I am implementing pagination on a webservice. My first thought was to use query params page and size, like Spring Data.

However, we are basing some of our design on the google webservice apis. I notice that they use pagination tokens, with each page result containing a nextPageToken. What are the advantages to using this approach? Changing data? What kind of info would be encoded in such a token?

Answer 1

When paginating with an offset, inserts and deletes into the data between your requests will cause rows to be skipped or included twice. If you are able to keep track in the token of what you returned previously (via a key or whatever), you can guarantee you don't return the same result twice, even when there are inserts/deletes between requests.

I'm a little uncertain of how you would encode a token, but for a single tables at least it seems that you could use the an encoded version of the primary key as a limit. "I just returned everything before key=200. Next time I'll only return things after 200." I guess this assumes a new item inserted between requests 1 and 2 will be given a key greater than existing keys.

https://use-the-index-luke.com/no-offset

Answer 2

One reason opaque strings are used for pagination tokens is so that you can change how pagination is implemented without breaking your clients. A query param like a page(I assume you mean page number) is transparent to your client and allows them to make assumptions about it.