Oh, I actually meant T5595.

The only workarounds I can come up with are:

Some followup in T7841 and elsewhere but this is substantially functional.

All typeaheads/tokenizers should still work correctly.

A big chunk of this is in HEAD. This should be where we are, roughly:

Quicksand navigation should probably add more requests to the existing DarkConsole on a page.

In D12444 I changed the default Differential query to use viewer(), instead of implicitly resolving the user's PHID. That means that when you "Edit Query", you get:

"Object" and "Global" rules don't punch through settings, so that rule shouldn't be the culprit unless I screwed something up with the implementation.

T7826 was a recent similar case where Herald is powerful enough to express something, it just isn't obvious how to build a ruleset.

We should definitely write a "useful things you can do with Herald" or "Herald Recipe Book" document at some point, there's a bunch of stuff that isn't very obvious until you know what the answer is.

Can you add a condition like:

Ah, yeah, if you're far behind HEAD you probably still had the 1 hour behavior. I'd expect that updating will improve things. Let us know if you run into anything else.

D12441 actually implements this beautiful, elegant control.

Very likely this is my fault then, I'll take another crack at scroll stabilization.

Yes, APC/OpCache maintain a cache of compiled code. In this case, a newer "DiffusionQuery.php" was calling into an old "ConduitAPIMethod.php". When you upgrade, make sure to restart afterward to clear the cache.

Second guess is that you have a large image on the page somewhere above the inline comment, and this is an interaction between image stabilization and maybe Differential content stabilization.

The issues themselves wouldn't cause a problem, but T7830 was a similar issue with UI/positional glitches related to the banner itself.

Try restarting apache (if using apache) or php-fpm (if using nginx + php-fpm) and let me know if the issue still reproduces?

Just seeing if I can go 2/2 here: are you an administrator on your install, with a persistent "unresolved setup warnings" banner showing? Specifically, do all pages which exhibit this behavior for you have the banner on them?

Should be up to date now.

We don't look for PID files or examine the process table because:

In fairness to the admins, we pushed a "deprecated Conduit methods called in the last 30 days" warning recently which often isn't easy to fix since you may have to chase people down or update things.

(I'm not sure if it's actually the problem, but it seems likely.)

I need to implement T6526 so I can award myself a MASTER SLEUTH badge.

Are you an administrator, with an "Unresolved Setup Issues..." banner showing on every page?

I cheated them into existence, but unified, browsable lists do work now:

I think I'm going to give up and use offset paging.

You can configure branch-level protection by allowing dangerous changes and then writing a Herald rule to protect specific branches. For example:

I don't think this makes sense separate from T7820.

That change adds some general instance administration documentation. It will go out the next time we push the cluster, usually Saturday morning if nothing comes up before then. I'll update this once it's live.

T7638 might be another approach -- roughly, a checkbox like:

To make this a bit more concrete:

I can repro so I'll take a look, I don't think I'm doing anything special (I have a mouse plugged in).

If a user joins the Policy project here and by doing this they get a "Member" badge for free

We should organize the documentation better, but here's the relevant part: