Thank you for testing. I think this looks good. In the one case that the old var is used it’s rendered as a string in pht so string assumption seems okay here.

One by one

thank you!

yarp

The underlying issue is way more contained now that rP7868ab3754fa is merged. I'm abandoning this revision to clean up my backlog.

Break long line into two

If we find 10 minutes, let's test this on every provider, as last time it gave us a little surprise on one of them

Oh neat, I see. Thanks!

Sounds like an outline-offset of a couple pixels or so would solve the general case, and individual elements that still look problematic could be revisited afterwards. I'm somewhat used to weird cut-offs and overlapping outlines and such myself, and I'd take that over missing outlines any day.

That is, what I get from setup issues at https://issuetracker.revi.xyz/config/issue/deprecations/ is to report it:

For completeness: Cf. T15808#17058

Also, phorge version and arcanist version is included in line 2.

@revi: Hi, could you please provide a list of steps to reproduce, if possible, and always exact Phorge version info? Thanks!

@revi: Hi, could you please provide a list of steps to reproduce, if possible, and always exact Phorge version info? Thanks!

Phorge saves CPU cycles

Or wrap a text with a background color - (something like What if I can change the background color here (and no italics)?)

To find it, I needed to put in loads of debugging. It seems to be an issue with PHP deferring the execution of the object destructor until after the Imbalanced AphrontWriteGuard check, even though the object was unset prior to the check.

Perhaps the real treasure was the friends you made along the way?

Also it's strange that the stack trace does not mention PhabricatorConduitAPIController

Thanks. Premising that I cannot reproduce the issue. I use arc diff and arc branches frequently without issues 🤔 Tested both with and without authentication.

Thanks again :) Giving right credits

For the sake of completeness: Please add mention of rPd519f75dfdee61bb109468aa708c47f53f3e5128. Thanks!

Use correct markup for comment about markup

Double-accept. This means you unlocked a rare pterodactyl1!1

Documentation is arrsum

Added minor tip.

Odd, I can't quite figure out how this would interfere with //Optional list<string>.//

I love that Phorge has this "Unbreak Now" btw

I'm somehow even more inclined to propose to base this feature on a new custom parser, like cowsay {{{ asd }}}

T15802: Regression: HTML entities rendered as plain text in config option descriptions

I've used git bisect and this regression was introuced after 4d12d014fd9ddc423ecbc6a61a92217d5aff19a4

Thanks everyone for the feedback! Looks like my initial assumptions are already uncommon, so I'm going to decline this ticket for the time being.
The counterproposal makes more sense to me on an 'upstream' level, indeed.

Abandoning per feedback in T15780 (note to myself: may want to consider deploying this in downstream)

Move additional check after existing checks

I would probably put this check after the New username is required. and also after the username validation, but that is still OK

If you followed the test plan and if this works, thanks

In D25601#17036, @valerio.bozzolan wrote:

Maybe better to expand the test plan to check user creation as well

You can set "visibility" for each URI already.

Example: an organization can have 100+ hosted repositories, both Subversion and git, and the highly-trusted-staff can be allowed to push directly in these hosted repositories, for example because they just rely on heavy post-commit review with Audit. And, they can rely on Differential just to share stubs. The pre-commit review with Differential has little to do with Diffusion here.

I visually mark this as "some discussion needed" in T15780

Premising that I quite understand the difference between Differential and Diffusion, I don't get why Differential should impact Diffusion here.

I cannot reproduce the above arc unit issues, locally.

• add hardening suggested by dear avivey
• add unit test for the new low-level function and its crash

Minor clarification.

It works. Thanks!!!