ThePythonicCow

There seems to be a bug in the vBulletin forum software (currently at Version 4.0.3) that iTulip uses. Occasionally a new post to a thread gets the wrong time stamp, typically sometime years in the past, causing the post to be misplaced near the top of its thread as an old post. Often at about the same time, one gets an "invalid security token error" when trying to post, which can sometimes be cleared by refreshing the page and trying the post again. That can be annoying if it causes one to lose the text of ones post.

It seems that one can actually track down when such incorrectly time stamped messages have been posted by

1. doing an "Advanced Search",
2. click on "Search Single Content Type",
3. find posts "A Year Ago" and "Older",
4. sort results by "Last Posting Date", "In Ascending Order",
5. show results as "Posts",
6. and click on "Search Now".

Examine results (approximately) #34 through #49 of the search results. These will all be replies dated in the first few months of 2003. Those timestamps are more or less all certainly incorrect. In each of the cases that I examined, when looking at the thread containing one of these 16 posts dated 2003, the thread was rather obviously running in the last year or two (2008 through 2010.)

A brief glance at the software bugs fixed or known in vBulletin (http://www.vbulletin.com/forum/conte...Existing-Issue) suggests to me that there have been a couple of possibly related reports, but that there has not been enough information provided to identify the issue. However I am not greatly familiar with vBulletin, so I could easily be quite wrong in this suggestion.

It seems to me that the 16 posts found by the above search could provide significantly more detail and evidence to the vBulletin software team, allowing more progress to be made on this bug. However I do not have access to, nor knowledge of, the internal iTulip data and fields that evidence this apparent corruption, so am not in a good position to file such a report. I would not be able to provide some of the most interesting evidence.

Certainly the particular frequency with which the problem occurs, say one in one thousand or one in ten thousand, and the fact that all the bogus time stamps (as found by this search anyway) were back dated to the early months of 2003 should provide some additional clues to the vBulletin developers.