Join Date: Jun 2014
Location: Northeast USA
Bikes: 2007 Specialized Sirrus
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
I've seen this problem with another vBulletin-based forum. It's not a bug, but a setting. The reason for both to occur is because the active user session times out fairly quickly. The session is refreshed each time you navigate somewhere new on the forum. The problem arises when you spend too long typing up a post, reading through a thread, or even just step away and come back later. At that time the session is ended and then reopened, resetting everything. If you didn't have "remember me" checked then the reset points would be obvious, as you would be logged out by the system every time it happened.
As far as I know, these are the options available to the forum administrators:
1) Inactivity/Cookie Based
Once a user has been inactive for a certain amount of time (the value of the cookie timeout option) all threads and forums are considered read. Individual threads are marked as read within a session via cookies. This option is how all versions of vBulletin before 3.5 functioned.
2) Database (no automatic forum marking)
This option uses the database to store thread and forum read times. This allows accurate read markers to be kept indefinitely. However, in order for a forum to be marked read when all threads are read, the user must view the list of threads for that forum. This option is more space and processor intensive than inactivity-based marking.
3) Database (automatic forum marking)
This option is the same as a previous option, but forums are automatically marked as read when the last new thread is read. This is the most usable option for end users, but most processor intensive.
We are probably on option #1
. The behavior that you're expecting utilizes option #3
. Personally, I'd like it if the forums switched to that, too. The administration likely shied away from it due to the note about it being the most processor-intensive. The forum I observed made the switch when an administrator had the time to monitor things, and he ultimately determined that the load wasn't bad, and left it enabled. Maybe we could try something similar?