The previous code path was buggy. We now do separate pointer update
checking for the cases where get_updates returns immediately vs. when
it returns from a callback.
(imported from commit f236a80cd0b94bc097dbd17f113d7a9d27368025)
For now, the new function, format_delayed_updates_response, just
calls format_updates_response.
(imported from commit dd332125fe0d47cb3990373f74e85e64604f58a3)
This seems more likely to be what people want when e.g. viewing all huddles.
Later we can get more clever and select the nearest huddle, or something.
(imported from commit 8ec1aa02c050dd25eb868b1e317d114743525c7b)
Our old selector here was picking up the floating recipient label, or
something. This was breaking narrow-to-all-huddles when on a stream message --
you get no selection and the code is sad.
(imported from commit d25fab03bd6d745df6d787c0b1b6452c8c539e32)
- s for stream or huddle
- S for subject
- p for all personals (still buggy)
We've eliminated the 'g' prefix entirely. And the old 'g-a' un-narrow sequence
has no new equivalent, but you can still use Esc.
(imported from commit 6bdaacce9639e3f749418e06622a317937e7d014)
Ideally this would be part of hiding zhome, but right now zhome/zfilt are
assumed to the tables themselves, and changing that seems unfortunately
invasive. And it's not crazy to think of the "loading controls" as a logically
separate thing that we might show/hide independently.
Longer term, we may want an indication in narrowed view that there could be
more messages on the server.
(imported from commit eb72d720da7c03f6f1378ae18ab6e973bf98247f)
The server will occationally return successfully with no data during
a restart. Previously, when we would get such responses, the page
would stop working because of JS errors.
(imported from commit 39b89907dc5ae9a9eb54ebf60d0069281e401786)
Reinstated from 65d00bf911231cb15d3a6713e64ed8dff28a613a . The server's
Markdown parser now treats newlines as line breaks, so we need this again.
(imported from commit 242ed53b7f1308f21a6e6a194e24c176545a079b)
This fixes a problem where the client would never stop asking for old
messages (bug introduced during rebase)
(imported from commit eef9f19e1e9982b1e0a954eb36a81e1b7ee5b564)
This is the format Nginx wants. It contains the following files concatenated
together in order:
app.humbughq.com.crt
startcom.pem
sub.class1.server.ca.pem
(imported from commit 6ed5679501f48729b4b4df19169042d5c0c37b5d)
This allows us to check whether the session that updated the pointer
is the same as a session that is doing a long poll to avoid sending
new pointer information when that information is coming from the same
session.
We still return from the long poll early, though, which is sub-optimal.
(imported from commit 7d4be0956f112eacefb7d198ea929957cd2b05e3)
The client may now optionally send its current pointer during
get_updates and the server will return the latest pointer if it
differs and was updated more recently by a different session.
(imported from commit e43b377d7dfb52f83cefb0b1003863d5407caf80)
Zev Benjamin
Keegan McAllister
Tim Abbott
Waseem Daher
Luke Faraone