We need a "Resolved: Ghosts" status.

For now I have a ugly idea: Change content in PhabricatorMailImplementationPHPMailerAdapter.php and add some lines to invoke another python class of my own.

See also T5462

Thank you! I read Describing Root Problems but I guess I missed some important points.

See Contributing Feature Requests, and use Ponder or Conpherence if you have general questions on how to do something in Phabricator.

If a linter could theoretically catch it (conventions, space, formatting, etc.), I think a linter should (in a perfect world) be catching it.

Do you bucket style guide violations with lint issues? While I obviously can't post internal reviews, I can give you a hand-waving indication that a non-zero percentage of comments are style-related things that aren't caught by linters.

Broadly, the technical stuff here is in fairly good shape now but I still don't truly understand what problem we're solving. From earlier, I dug this up:

I'm just going to merge this into T8250. I'm still unsure what's to become of this, but I think however that resolves will cover this.

This is effectively not a problem at HEAD, although I may make it a problem again soon. But I'm also not sure if anyone else has actually hit this, and it's a pain to fix, and even in the worst case it's less bad now because we pointer-events this. I'm just going to punt for now, let the dust settle on T12616, and we can revisit this if anyone's actually hitting it after all that.

Per T10563#221154, D17938 makes "reply" put comments in the right (threading-respecting) place and marks this complete.

(Offhand this sounds expected)

Bug reports require version information, and repeatable, detailed reproduction information. Please!

@epriestley I see, tried the same steps and it doesn't play directly within chrome. Thanks for investigating.

I can't play it in Chrome in any circumstance.

I can not play the .mov file in Chrome no matter what I try:

@chad The first video still doesn't play in Chrome (Version 58.0.3029.110 (64-bit); seems working in FF and safari now.

This is what I see in chrome (Version 58.0.3029.110 (64-bit))

Here's a screenshot of this working correctly for me in Safari:

@avivey I no longer run into this problem, please feel free to close this bug.

@epriestley not sure if the current phabricator install has been upgraded to lastest code that contains this fix yet, but the first video included in this bug which is a mov file still doesn't play. If I download it, it works, so the mov file itself is working.

While it would have saved me some time and some wondering if this kind of error message would have been in place I agree, it would ideally be in git. I'll see if there is anything about that on the git front.

We could say "checkout failed, you might have a bad hook" but I'd want to wait for more users to run into this first, and maybe try to pursue a fix in git with the upstream.

There doesn't appear to be any reasonable way for us to perform something like git checkout --skip-hooks, and I think we shouldn't even if we could.

Ha yeah that is interesting

Can you show me the output of these commands?

I can no longer reproduce the edit issue at HEAD.

In this screenshot:

In T5698#223992, @chad wrote:

Oh hai Cura uses Phabricator? I have a Lulzbot at home!

Oh hai Cura uses Phabricator? I have a Lulzbot at home!

On mobile, the hamburger menu icon is not aligned properly:

@Krinkle, @pablocarrillo I second your requests. Great suggestion. I have a large README.md that I would love to split into multiple documentation files, without losing all the formatting.

This no longer reproduces for me at HEAD of master.

I papered over this in the short term by restarting daemons for all instances;

We currently have three distinct selection states which I think are a little muddy. The first is the keyboard selection cursor, which shows what is selected with the keyboard:

@nemobis, can you walk me through the major differences you see between the proposal above and the support plans offered by other vendors like GitHub, GitLab, and Atlassian so I can make things more clear?

This is a productive environment, so I could not test to reproduce it and I don't know how can I reproduce it.

If I read this correctly, people would pay a certain amount regularly without any certainty or reasonable expectation to have any of the fixes or support they care about. Instead they'll need to rely on the resident haruspex to assess what they're embarking on.

Okay, but just to point out: having an apostrophe in one's name is not that exotic...

• O'Brien
• O'Connor
• D'Angelo
• O'Dowd
• O'Malley
• O'Hara
• D'Arcy
• D'Addario

So someone could quite conceivably have Chris.O'Dowd@BBC.co.uk or something like that.
Anyway, if anyone's looking for how to get around this in the future, an alternate solution to changing PhabricatorUserEmail.php would be to go into the DB phabricator_user, in table user_email, and change it directly in there, thus bypassing validation.
If you did want to change the code, though, look into http://php.net/manual/en/filter.examples.validation.php - FILTER_VALIDATE_EMAIL and FILTER_SANITIZE_EMAIL.

Per T12359, we aren't interested in supporting technically-deliverable-but-bizzarely-exotic addresses in the upstream. Feel free to fork to support this kind of address.

I've experienced the same error and it turns out that you need to give all privileges to all databases, not just phabricator. Hope it helps.