Sounds good! It's about what I expected, really.
(I did set up an experimental install on a whim, with the repository on-disk being (symlinked to) the root of the running install, and arcanist wasn't happy landing changes to it, probably for good reason heh.)

Uh, looking at line 112, the return status seems correct and expected. I had not seen that first. Thanks for fixing the constructor.

Basically, premising that the constructor of newFromFutures() seems legit, it's not clear to me why its return result should make sense and be compatible with the expected one. If somebody can clarify this, maybe it's at least something more.

In general the reporter should really find a small way to reproduce their problem, also because among all the things written here (https://we.phorge.it/book/contrib/article/bug_reports/), everybody may agree with the first point:

So this patch which changes (and hopefully also fixes) definitely broken code will be blocked for good due to lack of a test plan (which I won't be able to figure out)?
Or does anyone can come up with any other option?

Going to abandon for now as I'm not convinced either that this is something worth to change only to make PHPStan happier.
Note that https://we.phorge.it/book/libphutil/function/assert_instances_of/ might be an option when/if I want to look into this again at some future point.

Hello boz,

Slam-accept

Dear arp, please clone and hack a lot on Phorge :) Thanks and feel free to share any question

What's the status of this one? Just curious if we should fork and hack around it ourselves for now or if it's coming soon.

Don't create an empty title array when there is no $query_text

Uhhh you found the place where the empty array comes from! <3 I failed!

Taking the best from the two: what about assuming a default @link /book/group/best-document-ever-about-link - but still allowing custom @link to allow easy migrations?

https://we.phorge.it/T15911#19487

After looking again here:

Initialize key as an empty string as proposed by speck

What Valerio says: As $body can be null, make sure that the rest of this function's code deals with an empty string and not with null

Works locally; this makes sense to me (though I'm surprised it's in an Audit related source file).

Rebase; improve PhpDoc

I hope that something is better than nothing I think. Then we can move if we don't like it.

If Harbormaster is fine I better be fine too, dont want no problems with them for reals

Those were just for the looks, but this is of course personal.
I can remove them if you want.

even simpler

Ah! yes - undo ConpherenceViewController

(I don't get the comment about D25784) thanks for feedback

I guess D25784 shouldn't be mashed up in here thus requesting changes.

Uh!

I fail to reproduce a problem locally (after commenting out entire ArcanistBundleTestCase::testGitRepository due to a different issue I'm running into locally).

In D25807#21516, @valerio.bozzolan wrote:

P.S. try applying D25809 to check if the red unit test becomes green

@mturdus: Thanks for the patch! I tested this locally on a wiki page with a bunch of sections and on http://phorge.localhost/book/phorge/article/remarkup/ ; and on Differential's /D1 where it does not affect the ToC panel (good).

fix typo

Deprecated. See wojak_depressed macro instead - https://we.phorge.it/macro/view/14/

LOL don't take this too seriously but sometime this cursed thing makes sense