Make thread deletion detection smarter

[GoogleCodeExporter]

Threads that have been deleted [as opposed to singular posts] are not being 
properly tracked as such. Instead of being marked as deleted, the thread simply 
remains in the archive.

Reproduction: Create a thread on one of the archived boards, allow it to be 
archived by easymodo, then delete the thread. Wait for easymodo to check the 
board again, observe that the deleted thread is not marked in the archive as 
having been deleted.

Original issue reported on code.google.com by Kyl...@gmail.com on 18 Apr 2011 at 9:18

[GoogleCodeExporter]

Edit: Reproduction seems to not reproduce it.

Works as intended when following reproduction steps. Only breaks when a 
moderator deletes the thread, in which case it does not get properly tagged.

Example threads:
http://archive.easymodo.net/tg/thread/14635982
http://archive.easymodo.net/tg/thread/14636488

Original comment by Kyl...@gmail.com on 18 Apr 2011 at 9:55

[GoogleCodeExporter]

Marking a thread as deleted follows a dumb heuristic: if a thread disappears 
within the first hour of creation, then it's assumed it was forcefully deleted 
(either by a janitor/mod or by the OP). If not, it's assumed it died of old age.

By definition, all threads are deleted from 4chan, so obviously we need a way 
to draw the line. We *could* try to monitor the last few pages to infer which 
threads are dying from old age and which ones aren't, but that's really more 
trouble than it's worth, hence the simple 1h rule.

Original comment by eksopl on 18 Apr 2011 at 10:41

• Changed title: Make thread deletion detection smarter
• Changed state: Accepted
• Added labels: Priority-Low
• Removed labels: Priority-Medium

[GoogleCodeExporter]

It's a pretty minor issue, the only reason I even bring it up is because I'm 
looking for evidence of the crap one of the mods keeps pulling on /tg/. Perhaps 
instead of deletion within 1 hour, reduce it to deletion within 15-30 minutes? 
Most people tend to correct mistakes in their post almost immediately, whereas 
a moderator a bit to show up.

As I said, it's only an issue for me because the /tg/ mod has really started to 
become overbearing and it helps to have evidence when pointing this out to 
others. [And likewise, the archive would exonerate the mod if he *is* doing a 
good job.]

Original comment by Kyl...@gmail.com on 18 Apr 2011 at 11:25

[GoogleCodeExporter]

Reducing the time would just make your problem worse! I'd have to do the 
opposite, which is to increase the time in order to catch more deletions. But 
in case the board is going too fast (say, GET rush), more than one hour would 
mark too many threads as deleted.

Original comment by eksopl on 19 Apr 2011 at 1:10

[GoogleCodeExporter]

I marked this accepted, but I'm not really sure how to fix this.

We could keep a list of the threads that are in the last pages and mark a 
thread as deleted if it's gone and doesn't show up in that list, but that'd 
force people to crawl last pages more often. So I don't really like that 
solution.

I'll probably just make the time threshold user adjustable and settable by 
board.

It's an enhancement, in any case, not a defect.

Original comment by eksopl on 4 Jul 2011 at 5:00

• Added labels: Type-Enhancement
• Removed labels: Type-Defect

[GoogleCodeExporter]

Moved to Github.

https://github.com/eksopl/fuuka/issues/40

Original comment by eksopl on 23 Apr 2012 at 12:42

• Changed state: Done