Hell if I know
-
-
OFFSET must not be negative
Apparently…
-
So just use a positive offset, what's the problem?
-
-
@pie_flavor oh no not postgresql error messages
-
-
Can you post the socket.io request that is causing it? You can see in the WS tab of chrome. Here is a screenshot when scrolling down a topic.
topics.loadMore
is the call that loads more posts.
-
Also which topic does it happen in? The one I tested was fine so need to try in the same topic.
-
@barisu I was able to continue to do it while the tab was open (hence the recording of it) but I wasn't able to do it after the page was reloaded. Guess it was a temporary thing. Still, I get one hell of a lot of temporary things. NodeBB does not really like it when it changes networks.
-
@pie_flavor In Discourse times these were called "temporal bugs", I think named after something Jeff said
-
@hungrier said in Hell if I know:
@pie_flavor In Discourse times these were called "temporal bugs", I think named after something Jeff said
Um, actually, I can see why you'd think that, but "temporal bug" was 's word for "regression."
-
@boomzilla said in Hell if I know:
@hungrier said in Hell if I know:
@pie_flavor In Discourse times these were called "temporal bugs", I think named after something Jeff said
Um, actually, I can see why you'd think that, but "temporal bug" was 's word for "regression."
Maybe, but IIRC we adopted it to describe any bug that was either temporary or intermittent.
-
@hungrier said in Hell if I know:
@boomzilla said in Hell if I know:
@hungrier said in Hell if I know:
@pie_flavor In Discourse times these were called "temporal bugs", I think named after something Jeff said
Um, actually, I can see why you'd think that, but "temporal bug" was 's word for "regression."
Maybe, but IIRC we adopted it to describe any bug that was either temporary or intermittent.
Like I said, I can see why you'd think that.
-
@pie_flavor said in Hell if I know:
but I wasn't able to do it after the page was reloaded
NodeBB does not really like it when it changes networks.
It's almost certainly the weird state that happens when the websocket hasn't reconnected right. I wish there was a way that we could get that layer to stop trying to struggle on, throw away that broken connection object, and reconnect correctly from first principles, but that seems to be a really difficult thing. (I guess because someone in the browser's security team has decided that the local socket address is also a critical part of the page location? Or something of that order of annoying.) I guess part of the reason this is awkward is because most testing/development labs don't have the setup for switching computers back and forth between networks…
IOW, it's the usual goddamn bug with slightly different consequences.
-
@dkf said in Hell if I know:
I guess part of the reason this is awkward is because most testing/development labs don't have the setup for switching computers back and forth between networks…
What? Like setting up a few WiFi hotspots?
-
@DogsB said in Hell if I know:
What? Like setting up a few WiFi hotspots?
Yes. Exactly like that, since they still need to have connectivity and most sensible network admins are extremely paranoid about extra wifi hotspots turning up (since consumer gear is often a bit misconfigured; the worst is when it decides to be an alternate DHCP server for everything on both sides of the wifi bridge).
-
@dkf said in Hell if I know:
the worst is when it decides to be an alternate DHCP server for everything on both sides of the wifi bridge
That was incredibly annoying when that happened to me. Turns out Microsoft's DHCP server is very shy around other DHCP servers for some reason. Or, the default configuration is set to be really nice.
Fun times will slowly be had if the server is rebooted with a rogue DHCP server on the net...
-
@dkf said in Hell if I know:
@DogsB said in Hell if I know:
What? Like setting up a few WiFi hotspots?
Yes. Exactly like that, since they still need to have connectivity....
Would a VM with multiple fake network interfaces work? "Unplug" one, "plug in" another, and go from there? Maybe that's too much of a change, but something along those lines might work.
-
@Parody said in Hell if I know:
Would a VM with multiple fake network interfaces work?
Perhaps. It does require someone to realise that there's a problem with this sort of thing in the first place and that that's the way to fix it. (I've seen good things with websockets when the network goes down and then comes back up with the same address; that's obviously not the problem here…)