I might make some more tweaks to the UI, but this looks like it's in better shape now.

The actual replacement is Authorization: token <token>, I believe:

I just wanted to hold it across the release cut, it's headed to master shortly.

Both of these hosts restarted cleanly.

AWS is also rebooting web007.

See https://discourse.phabricator-community.org/t/herald-rule-not-adding-auditors-to-commits-following-upgrade/3505 for one bug with this.

Sounds good. It felt slightly odd to me at first, too, but I think I also got used to it.

This appears to be working. T13481 fixed one straggling issue with changes to remote handling in very old versions of Git but nothing else has cropped up.

The empty space means "this is a normal commit with no special audit state", and the column collapses if no commits have issues (see this task for an example). That seems fairly reasonable to me?

I think D20905 is as good as we're going to get.

The underlying code is still a bit shaky, so let me know if you catch other issues.

In DifferentialHunkParser->generateVisibleLinesMask(), we pass a 0-based array in and get a 1-based ("line number") array out. The loop condition stops us from reaching the last line of the 0-based array.

We have no open customer requests for this and it's very complex, so I don't currently plan to implement it.

I removed the strikethrough in T13482, which mostly moots this.

Remaining work:

D20947 does not implement "Author's packages" as a "Commit Content" field, nor as a "Commit Content (Hook)" field. The reason for this is that getting the modern authorPHID in both cases is somewhat complicated.

The logic here appears to be that gc.auto is set to some value (by default: 6,700). If the number of loose objects exceeds this threshold (technically, if the number of loose objects in objects/17/ is more than 1/256th of this value), it triggers a repack (in a comment, git repack -d -l).

See PHI1613, where an install hit this warning (and resolved it by running git prune):

The issue in PHI1605 resolved itself without apparent intervention, presumably as a result of changes on the CircleCI side. I can't find any release notes to shed any light on things, but this is no longer time-sensitive.

See also https://discourse.phabricator-community.org/t/storage-upgrade-failed-class-phutilauthadapter-not-defined/3468.

See T13478 for followup.

I'm also still able to reproduce this specific variant of things, so it looks like I was too optimistic about T13462 covering this in at least some cases.

A similar issue occurs when the "Visible To" control wraps with spaces enabled. To reproduce this...

This should be fixed by D20943. Note that "Mute" in this context mutes notifications about edits to the object (e.g. "Alice renamed rule Hxxx from X to Y."), not notifications sent by the rule itself.

This went through cleanly.

These changes seem to have stuck.

I thought this was some kind of complicated mess with the regex on line 420, but it's actually an issue with this:

(Since a full implementation here will imply that removing a custom field from configuration may "unmention" an arbitrarily large number of relationships, and we can't easily do that inline or at display time, the ultimately implementation would likely include a more robust version of this script.)

In PHI1602, I proposed a script to "reset" mentions on an object. The use case this serves is:

This page intentionally has a no-Javascript, no-drag-and-drop upload form for users who use obscure Linux window managers that don't support drag and drop.