« first day (1469 days earlier)      last day (55 days later) » 

7:27 AM
FYI, suggested edits on Metas can take a long, long time to be reviewed ... — Glorfindel ♦ Jun 2 at 11:59
@Glorfindel Certainly, it can be quite long - reviewing on per-site-metas isn't that active. If several edits are suggested at the same time, the red dot might speed things a bit. (On the sites which I frequent, the longest time I've seen was several months.) OTOH, this certainly isn't something that could be considered urgent - it's almost two years since your bug report. — Martin Jun 2 at 13:38
 
 
4 hours later…
11:56 AM
ah, nice
 
12:44 PM
8
A: An edit missing from PostHistory in SEDE

reneWell, this is no fun. The PostHistory table is fine. All the records are there. Notice how UserId is no longer a magic link, User Link. Having the User Link magic column magically makes rows go up in smoke. I tried to replace the User Link with the somewhat less magical manual link build option ...

Great ... now I'm wondering how many queries are affected by this ... — Glorfindel ♦ 17 hours ago
@Glorfindel we can't know. We don't have access to stored queries except scraping. I suspect the problem is either in QueryRunner or in ParsedQuery but none of the fail cases I imagine would lead to the query returning less rows. — rene 17 hours ago
@rene Thanks for posting the answer. I suppose that the right thing to do would be to accept your answer and consider this question resolved.
You have mentioned that this reveals that there is some other bug - but a separate question for that one would probably be better. (I suppose you or somebody experienced might be able to post a reasonable bug report concerning the unexpected behavior of magic links.)
 
 
3 hours later…
3:17 PM
Is it possible to detect for older duplicates with edited targets that are unanswered and thus redirect to the wrong questions?
Specifically, questions closed with the older duplicate reason that resulted in the target getting edited into the question by the Community user (reason ID 1 not 101), which have no answers and were originally closed with a single target, and where the automatically edited target was subsequently edited by a human.
When browsing to such questions, anonymous users will be redirected to the original target, rather than the one which was later edited (or not at all if there are now multiple).
 
Do you have one example post, somewhere? Then you can check PostHistory for that post to see if you can deduct that state from the info you got.
 
On the other hand, for newer duplicates, when a user with privileges edits the duplicate list, anonymous users are redirected to the new target (or not redirected if there are multiple targets).
@rene I couldn't find an example post. It's something I came up with as I was reading this answer of mine and the comments on the question directed me to this SO question, which is where I came up with the idea.
 

« first day (1469 days earlier)      last day (55 days later) »