Page MenuHomePhabricator

Plans: 2018 Week 11 Bonus Content
Closed, ResolvedPublic

Description

See PHI406. An install would like configurable favicons for development/staging/production installs.

See PHI433. An install would like to improve some of the messaging around drafts.

See PHI431. An install would like to support a nuanced workflow in review actions that probably doesn't make sense as a default upstream workflow but may mesh with T731 and which we may be able to provide some support for.

See PHI427. An install would like a way to render a short Doorkeeper external object reference. Update: this is hard because Doorkeeper uses the remote object reference as a cache key in both client and server workflows, and can't currently distinguish between different views of the same object. Not insurmountable but should actually get proper testing.

See PHI442. See T12994. This is probably worth resolving since it has cropped up again. Guidance for optimizer_search_depth=0 may also be warranted.

See PHI451. An ancient inline Javascript handler in DarkConsole violates the CSP header.

See PHI467. An install is hitting T8845 with L3 in a header.

See PHI364. An install encountered a held lock issue in a cluster which resolved itself immediately when I appeared onsite. It seemed plausible that this was because read synchronization does not put a timeout on git fetch, so if git fetch hangs for some network-related reason the lock could be held indefinitely.

See https://discourse.phabricator-community.org/t/desktop-only-notifications-mode-is-broken/1234/1. The "Desktop Only" notification mode seems to not be functioning properly.

See PHI461. An install would like slugs in project.search.

See PHI454. An install identified an opportunity to improve a joke in the API.

See PHI430. An install has identified a reproducible case where a policy exception causes a hard fatal instead of a graceful policy error.

Event Timeline

epriestley triaged this task as Normal priority.Mar 9 2018, 4:03 PM
epriestley created this task.

While testing D19215 I ran into this behavior in Asana:

Screen Shot 2018-03-13 at 11.12.08 AM.png (228×640 px, 36 KB)

Those buttons just fill in the literal text written on the button, almost. They've been programmed very carefully -- for example, "Quick question..." fills Quick question (no trailing space) while "I can start on this by..." fills I can start on this by (with a trailing space), probably because the former phrase likely needs punctuation (like : or ,) while the latter does not.

the quick replies only show when someone has assigned you a task that you have yet to comment on, with the goal to encourage folks to acknowledge their tasks :slight_smile:
https://community.asana.com/t/new-in-asana-quick-replies/6910

This feature shipped in August 2017, apparently can not be customized or disabled, and is utterly mystifying to me.

It appears that MYSQLI_ASYNC depends mostly on mysqlnd ("MySQL Native Driver"). This has been the default since PHP 5.3 was released in 200 BC, but Ubuntu, at least, unbundles it and you can still install php5-mysql (non-native driver). Compounding this, we currently recommend installs install php5-mysql (bad), rather than php5-msyqlnd (good).

It looks like apt-get install php5-mysqlnd just works correctly and replaces the nonnative driver with the native driver, so that's likely easy.

I also found this gem:

"Recommend you work on speeding up each query. Parallel execution in PHP ... is mostly a myth."
https://dba.stackexchange.com/a/111884

See PHI364. An install encountered a held lock issue in a cluster which resolved itself immediately when I appeared onsite.

I'm still expecting to fix this today but I'm going to roll it forward until after the release cut since the install in question isn't on a bleeding edge version right now and it wouldn't hurt to let this change soak for more than 5 minutes before it goes to stable.

See PHI431. An install would like to support a nuanced workflow in review actions...

This morphed into a bit of a bigger change and isn't fully resolved yet.

See PHI430. ... a policy exception causes a hard fatal instead of a graceful policy error.

This is improved in master but continued in T13088.


Everything else here should be resolved in master and promoting to stable shortly.