MySQL查询结果偏移量与新结果可能的限制？

Question

I'm working on an application that uses a scroll load system of 20 or so results loading into a feed at a time as you scroll. This feed consists of constantly added user generated content. Which means the result set can change per query that is offset by X.

So let's say we load 20 results, then scroll, another 20, and then before scrolling more to load the next 20, another user has uploaded a new piece of content, which effectively would present a duplicate in the next set of 20 results for the feed because we're using OFFSET to get additional results, and the total result set is getting shifted by 1 with this addition of new content that falls into the conditions of the query.

What is the best and most efficient way around this? We've dabbled with using the id of a row in a where condition to prevent duplicate results, and only using limit without offset for new results fetched.. so we could do WHERE id < 170 LIMIT 20, WHERE id < 150 LIMIT 20, WHERE id < 130 LIMIT 20, etc.. to control and prevent duplciates... HOWEVER, this does not work in every possible scenario as our result sets aren't always ordered with the id column ordered by DESC..

Soo.. what other options are there?..

Answer 1

You might want to try a database solution. On the initial request, create and populate a table. Use that table for the feed.

Make sure the tablename starts with something consistent, like TableToFeedMyApp, and ends with something guaranteed to make it unique. Then set up a scheduled job to find and drop all these tables that were created earlier than whatever you deem to be a certain interval.

Answer 2

Why are you using the where clause instead of limit with the offset option? Limit can take two arguments. The offset argument seems to do exactly what you want. For instance:

limit 100, 20

Takes 20 rows starting at the 101st row. Then:

limit 120, 20

Takes 20 rows starting at the 121st row. (The offsets start at 0 rather than 1 in MySQL counting.)

The one enhancement that you need to make is to ensure that the sort order for the records is stable . A stable sort is, essentially, one where there are no sort keys with the same value. To make this happen, just make the id column the last column in the sort. It is unique, so if there are any duplicate sort keys, then the addition of the id makes the sort stable.