Uh! I like this.

Probably OK now

Interesting triage. In legacy upstream this was a "feature request 🌈" - https://secure.phabricator.com/T12438

Super-useful comment:

I don't know if the file is really related, but PhabricatorDocumentRenderingEngine can be seen in scene while surfing a Diffusion repository and playing with View Options.

Relevant old upstream comment (from an unrelated task) that describes this problem as "ghosts in the UI":

adjust strings as suggested, also, renamed classes from "subscriber" to "subscribers"

Help welcome.

PhabricatorFileTransform: add null check

Thanks again. Generally super-useful.

Uhm. Incredibly, it seems to work. I have good feelings.

In D25636#17842, @aklapper wrote:

Ideally this would be done via validateTransactions using ->setFormErrors($errors) on a PHUIObjectBoxView but all seems quite broken for badges anyway, see T15827.

Damn celerity. Thanks.

Damn celerity

(Premising that this is an answer to your question. Please move this as comment to your question.)

Nice but please smaller :D

Thanks! I was able to test visiting a profile image, and inspecting phlog() in the webserver log.

Thanks. I was able to test that the thing is just a string, not an object with a toString(), so no nuclear implosions

(If it does not work you may still need to call .toString() before doing this)

For some reasons I cannot cleanly apply the patch anymore.

Just because we can, right? Or any real concern?

✅ Number.prototype.toString() supported by all browser webs https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Number/toString

This fix was super-interesting. Thanks again for the report.

Thanks again. My only concern is that, if somebody has a 1 trillion file(s) then the patch may lock the involved tables for an unexpected long amount of time, and you cannot really undo, without causing a massive implicit rollback.

It worked in line 109, it will work in line 154. Tested.

Cover more cases that require login:

Wow you are fast. I was going to create this :D I can reproduce

Commanding to set as abandoned. Thanks again! Well done for your original troubleshooting.

git rebase master

Just as clarification for myself (maybe already obvious to others):

Thanks again Roberto.

My last comment is also because the W3C does not allow to start an identifier with a number:

Please evaluate extended names:

Unrelated but, thanks to your patch, we catched a possible micro-optimization. Here that unrelated change:

Thanks again! This looks perfect in the current state of PholioMockEditor, since it's not an EditEngine and since this is not an appropriate moment to learn all the differences between an "Editor" and an "EditEngine" and/or it's not an appropriate moment to do that kind of migration here now (if ever needed).

Maybe "feature" but note the left sidebar, is not anymore connected to the end of the viewport:

Sharing some comments

fix abstract function visibility, add more PHPDoc

Setting high priority since I'm scared :D Feel free to re-tune

I was not aware that we were executing critical business code inside __destruct() automagic calls. Wow!

In D25626#17531, @aklapper wrote:

This seems to also fix http://phorge.localhost/feed/ so it does not only show me about setting high priority on a new task, but instead shows that the task was created? That would be a nice side effect too IMO.

Maybe relevant, from my Postfix logs: