Question Hot Thread Icons

Good idea, any coders out there willing to try this out?

sozzled wrote:

SELECT thread, SUM(hits), subject, catid, id FROM #__fb_messages GROUP BY thread ORDER BY total_views DESC LIMIT 0,10;

Sorry but that would not work. This query would calculate hits for every thread on your system.

First off that information already exists in the messages table, second the absolute number is not really any good for that. You need to calculate the number of views over a time period (e.g. the last 7 days) and optimally you would want to compare it to the system average of views and pick ones that are significant outliers. For example if the system average for views per thread per day is 10 and you have a thread that received 50 or more, you would want to mark it as hot.

If 3 months later that same thread only gets 5 views per day, it would no longer show as hot.

Now - all of that information is currently no collected by the forum. It is a great idea and I have some thoughts on how to make this happen while also making it scale for any size site, without the need for extra queries.
This wont happen for K1.6 though. We need another data model change that is already done in the Kunena 2.0 trunk plus an extra history table that stores views over time.

Great idea though and definitely something I would like to add to a future release.

Yes, I agree, fxstein, and the SQL query I gave above was only intended to be used as an example. Yes, it's necessary to put a date range into the query to isolate only, say, the past week or month or whatever timeframe is more appropriate for each user's purposes.

It's an interesting request and one that, I believe, would be useful to the project but, as you say, it won't happen at this present time. Something for the future and something for others to experiment with.

Instead of hits, I would use replies when deciding if the topic is hot or not. If there have been a lot of replies in X last days, topic would be hot. We already store that information and currently we are also reading that data anyway..