We should, yes. Until this was reported I wasn't aware of this issue and believed we had comparable support across versions.
- Queries
- All Stories
- Search
- Advanced Search
- Transactions
- Transaction Logs
Advanced Search
Feb 26 2015
I think this is a problem with logged-out users, anonymous access, and arc patch.
The way the logic actually works, I think the fix is really in PhabricatorAuthLoginController near line 117:
We should enforce uniqueness. The fix is probably something like:
The reason for this logic is so that you can ^C twice to get out if a daemon is hung and you're running it in the foreground. This avoids the need to bring up a second terminal, figure out the PID of the daemon running in the first terminal, kill it, hope you got the PID right, etc., or abandon the first terminal entirely.
Specifically, the driving use case is:
My expectation is that this row is always visible and there is no down arrow.
Resolved by D11889.
There's now a host prune command to destroy backups that are out of the retention period.
My take on this is that we should build this UI first:
Feb 25 2015
Oh, there's a bit of discussion in T5364. I think there are two ways to build the connection stuff:
I can contextualize this a bit, I think I had some ramblings elsewhere. Let me see if I can dig them up.
It doesn't exist and is very hard to build :)
We probably need to flag these images as builtins, adjust their policy settings at runtime, and then write a migration script to purge them so they regenerate properly.
Oh, sorry, you're right, it's the default image.
These are probably very old projects with very old images from before all the policy stuff, let me see..
One issue with this as it stands today is that it's hard to get the file to be marked as cacheable. Some discussion in T6007. Without the flag, the logo will reload/flicker on every page load.
T7376 added some more information here and would let us produce "You are out of date / your junk is full of local custom code which we don't support" messages.
send the entire document tree to the client
Okay, I don't quite understand your network setup then.
Ah, I wasn't sure if he was you or not. He asked a similar question in IRC and left before I had chance to answer.
Specifically, this isn't causing any problems yet (hg locate still works in modern hg at the time of writing), but at some point hg locate is presumably going to vanish and then we will have a problem.
Was this the same issue, or someone else? The paste link expired:
Sorry, my fault -- I only gave the VCS user sudo access to the git binaries when I set things up. I'll update the config when I have a chance.
Narratively within the system, these are still "in the queue", they just have a lease on them. They don't get taken out of the queue until they are archived.
If you also want https://phab.local-domain.com to work for users (for example, one URL when VPN'd and a different URL when not VPN'd), add it to phabricator.allowed-uris. We don't recommend this kind of setup because there are a lot of places where we can't generate a valid link if the install has different locations. For example, in email, we have to pick one of the domains when generating a link to a revision, and we can't know whether the user will be VPN'd or not when they read their mail.
From the log, it looks like brining the daemons online worked fine, and they hit that error about 8 hours earlier (maybe a one-time issue on SendGrid's side). Am I misunderstanding? I'm looking at the timestamps -- 25 Feb 15:00 vs 25 Feb 23:00.
(In the Phacility case, the 1x is about 2KB and the 2x is about 3KB. Clearly not worth special casing to save 1KB, once, on non-retina displays.)
We should be able to use one image and just background-size-x/y it to the right dimensions. On retina it will have 2x resolution properly, on normal displays it will be "too big", but that's fine (they won't be much bigger, and they'll get cached).
(I haven't seen this, but I'm almost always looking at the console while updating the box.)
The next time you see this, can you check IRC and make sure phabot didn't leave recently? I pushed a bit before this was filed, and this might be push-related.
Feb 24 2015
Seems to be working in the cluster, too.
This stuff seems to be working (daemons restarted cleanly; autoscale worked; queue flushed; no errors) so I'm going to roll it to the cluster.
My tentative fix here was:
I guess you'd probably need to override background-image-position-x/y and such too if it's currently using a sprite.
I think regenerating the map is really bad in the long run. Particularly, it means the sprite map and resource map will be dirty in the working copy, so git pull will fail the next time we update them and we'll get a bunch of instances that can't upgrade. Another issue is that Phacility users could not customize their instance headers, because they all share a single map. And they need optipng installed. And the daemons need write access to the source code itself if we're going to do this in the background.
Nah, I've got it in a branch.
It is possible that this is caused by a combination of:
Feb 23 2015
Remaining stuff here:
Feb 22 2015
I think 2-3 concentric circles of eldritch runes should appear and rotate in alternating directions. As they gain speed, the browser should emit a bass hum and begin to shake. This will help users understand the interaction better.
A few minor things came up, but instances are now good to go for "real work".
There's now a host swap command which automates most of the mechanics of this. Needs a little more work but this is significantly less error-prone now.
Feb 21 2015
In practice, the trial period exists: just sign up and disable your instance before the first invoice, or don't ever pay us. You don't need to give us a credit card or pay anything to sign up.
I think this should be fixed now, let me know if you see it again.