4
\$\begingroup\$

While poking around on my profile page I noticed that the "folded" versions of rep events don't contain the proper number of individual events once unfolded. It looks like the methods used to filter the events by time isn't uniform.

folded - 5 and 4 events, respectively: folded: 5 and 4 events

unfolded - 9 duplicated events each: unfolded: 9 duplicated events each

\$\endgroup\$
  • 2
    \$\begingroup\$ It looks like this might have been a temporary display bug. I don't currently see this on your reputation page; I see the correct 9 events that total to +45 under a single unfold-control, with no duplication or splitting like in the screenshot. Do you still see this bug when you look now? \$\endgroup\$ – SevenSidedDie Feb 22 '17 at 5:04
  • \$\begingroup\$ @SevenSidedDie No, I don't. It's possible that as the code used to filter events into these groups condenses the groups as time passes, thus hiding the problem when it is sufficiently far enough into the past. Maybe a temporary bug, though... I'll check back in next time I'm inundated with rep events :) \$\endgroup\$ – Conduit Feb 23 '17 at 18:21
  • \$\begingroup\$ I don't think condensing is a thing it does — that seems an unnecessarily complicated UI choice and not something I've noticed before anyway. More likely that yesterday's code base contained a display bug that was fixed by the end of the day. :) In any case I've marked the bug as closed (the “no longer happens” reason); thanks for the report! If it reoccurs though, add more info and comment or flag and a mod can reopen it. \$\endgroup\$ – SevenSidedDie Feb 23 '17 at 18:23
  • \$\begingroup\$ @SevenSidedDie Only a few lines of code if they're using the time functions of moment.js - the library that makes times look like "1 hour ago", "yesterday", etc. - there. Suppose it's quite possible that fell under the umbrella of the display bug, though. \$\endgroup\$ – Conduit Feb 23 '17 at 18:26
  • \$\begingroup\$ I expect they have some more complex code around the twisty and collecting the lines in that report though. Note that it's not just filtering that's in error: in the screenshot, all the entries under “5 events” and “4 events” match, so somehow they failed to collect together those 9 events correctly at the twisty/header level, but they correctly showed the full 9 events under each duplicate header for the one post. Some kind of matching bug in the header code. \$\endgroup\$ – SevenSidedDie Feb 23 '17 at 18:30
  • 1
    \$\begingroup\$ I've removed the [status-norepro] — it appears (see answer and its link) that it's not unable to be reproduced, just hard to reproduce due to lack of control over the triggering conditions. \$\endgroup\$ – SevenSidedDie Jul 28 '17 at 17:58
7
\$\begingroup\$

This problem has been reported on Meta.SE, and can be reproduced by generating a sufficiently large (~10) number of rep events over a time period of ~1 hour.

The bug has not been fixed as of July 2017.

\$\endgroup\$
  • 1
    \$\begingroup\$ Are you still seeing this bug though? \$\endgroup\$ – SevenSidedDie Jul 25 '17 at 17:49
  • \$\begingroup\$ Ah, I see what's going on: it's a context sensitive bug, so though you're not seeing it now, it's just because the data has changed, not because the underlying bug is gone. \$\endgroup\$ – SevenSidedDie Jul 25 '17 at 18:01
  • 1
    \$\begingroup\$ @SevenSidedDie - Yep. After a time the events become clustered properly, so the problem is only manifest for a short period after the events are first created. \$\endgroup\$ – Conduit Jul 25 '17 at 18:06
  • \$\begingroup\$ The question's tagged norepo, so if you can figure out a way to reliably trigger it, including that here might be helpful to the devs. \$\endgroup\$ – the dark wanderer Jul 28 '17 at 2:52
  • 1
    \$\begingroup\$ @thedarkwanderer - Unfortunately it can't be reliably triggered by a user since we can't generate rep events "dishonestly" :). I can attest it is quite easy to reproduce; just need a sufficiently large number of rep events occurring over a period of an hour or so. Added to the answer body. \$\endgroup\$ – Conduit Jul 28 '17 at 15:39

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .