I heard you. :D

In T7887#108441, @btrahan wrote:

Still very interested in any additional color on the problem here - e.g. examples of useless notifications no one wants to hear about and that you have no way of silencing to begin with

Unless @chad has an idea on how this UI could work, I think this is a wontfix.

Yeah, reasonable enough. Sometimes I'm just feeling picky and don't want the extra formatting on it though.

I usually just do backticks. D1234

Where this seems to get me the most is on "close" actions: a diff is closed, an audit is closed, a task is closed, etc.

Mostly, we'd rather improve Herald and Notification Preferences than build and maintain a new system. We want to make sure the right issue is being resolved.

Can you describe better for us the actual problem the user has, this task seems like just a feature request. Between Herald and the Notification options in Settings, there should be enough options here to improve the quality of all the notifications you receive.

I ran into this need when trying to cover all of my bases for an extended maintenance notification. We don't have an email list that definitely covers 'all users of phabricator' that isn't absurdly over-broad (like 'all employees'). I suspect most corporate users will run into a similar convoluted problem at some point.

Probably don't need the dialog at all IMO.

Tokens is T7468

Hmm yeah, that's a good point.

Sort of interesting use case, though I don't know we have the manpower of a Slack to pursue it (this year). I wonder if you could fake something with "work hours" where you want to receive notifications in both places.

Note that you always receive a notification. Going to Notifications will always show you everything.

My thoughts on this. (Not that I'm so important, but I'll leave them here rather than T7702)
My suggestion in there was to only send an email if the notification was unread for a set period of time and to not mark notifications read just because an email was sent.

Boom. Firewall. I'll stop spamming you now.

Well, it still says disconnected in my notification panel even though I see the notifications being added to the log and I don't have an unresolved setup issue, but at least it installed correctly. I'll keep debugging around and see if I can figure out why. Thanks for the help.

Ok, so I just removed node, and then added a symlink from nodejs to node and it installed correctly. Not entirely sure what node was or why it was there. Perhaps in your script you might look to use nodejs first and then node if not there? Or maybe I just have a screwed up system.

Hmmmm, is there a difference between node and nodejs? I can't seem to get a version from node.

We haven't seen this problem with npm before, so I can only offer vague advice -- are your node and npm up to date? Does installing other modules with npm work? You may have better luck filing against npm.

At least at one point in time, registering anything on beforeunload would disable the "bfcache":

Technically speaking:

Confirmed.

I intend to keep the colors/counts in the menu (plus the count in the page title) in sync across windows eventually. Generally, more stuff will probably sync in the future than it does now, although I don't currently plan to ever make transactions sync in the general case (but we'll see).

Yea, when writing this I wasn't really sure what term to use, lets see what Evan says :P

(I would like this to work like you state regardless, it's just as I expect it to work currently).

I think term "live update" is confusing this here. Maybe a better way, at least as I understand it, is only Conpherence's UI is kept in sync. Notifications are not kept in sync with the server, they're just sent and the individual tab deals with it. Changes are not pushed back up and then back down again. Though I am just a lowly Jr Support Intern, so maybe I am wrong about the current state of Aphlict. haha

Notification bubbles do live update if you use aphlict, well, they live increment, they don't live clear themselves which is the issue im getting at.

That's not exactly what I mean. The only app with live updating UI is Conpherence. No other UI auto updates. Notification bubbles are not live updating, they are just notifications that something happened. You always have to reload to see the changes.

Aphlict notfications as well

Conpherence is the only realtime updating application.

You, sir, are a gentlemen and a scholar.

The goal is to remove the need, not to remove the actual email, just to clarify. If I can set up local filters to hide all Phabricator email except [Maniphest][Created] then I see that as the perfect environment for me. People can adjust local rules to suit their needs, and the ones that want to completely disconnect still get a rich experience.

I agree for the typical developer use-case. But currently people reporting bugs/suggesting features use maniphest as well and for them the email notification is crucial. So always sending them is probably best for now.

This would have to be a setting, and generally not super fans of settings here. Mostly because if we're going to be automagical, would prefer the user opt into it.

I like the idea, but i think it's a tricky thing to get right. On most sites that do this i always had occasional problems not getting email notifications when opening/closing the site repeatedly during the day (which also is a common thing for me with tools like pahbricator).
I think notifying a user via email if he doesen't act on a web notification in a certain amount of time would be best. But that might make it way more tricky to implement.

See T6944: Try to create the pid directory in Aphlict.

I got an error about trying to write to /var/tmp/aphlict/pid, which did not exist previously as far as I know. P1699 contains the full trace I got.

The update went pretty well, and it seems to work just fine!

Wow I just saw notifications on my iPhone

This is really awesome gentlemen.

It seems like the root issue was resolved through dashboards / hacking in notifications.

Body of the email:

*** committed rRK3477: default-deferred excel template for backward compatibility in context of T52.

Is the first section of the mail just like:

Just to clarify the case further:

I can't reproduce it anymore after restarting of the server. Sorry for the trouble.

Can you show me the X-Phabricator-Mail-Tags header for the relevant message? You may need to "Show Raw Message" or similar in your client to see it.

[[ ]] is part of Remarkup, so it has to be typed in directly by the comment author (making the URL appear normal in the comment if they didn't name the link). We don't convert these strings.

I can't reproduce this. Can you make a comment on this task which you believe reproduces the issue?

Was thinking on this today -- most of the need I would have for this is the maintenance case of "all users need to know this service will be offline on x date". People ignore emails so point of presence of this kind of thing is nice :)

Will revisit with Phacility down the road. Even if we implemented, it seems unlikely users would be able to use it on local installs.

Interesting, understandable. I'm curious though why your own domain (sending/receiving) isn't automatically whitelisted.

I've looked at it when it first came out, so maybe it was streamlined a bit since.