I was not able to reproduce obviously badly broken behavior from these
logic bugs, but after the renaming of message_viewport helpers in the
last few commits, it's clear that this logic was trying to check if
we're actually at the start/end of the possibly message feed, not just
the rendered portion, and doing so incorrectly.
The previous logic for both scrolling down and using pagedown would
incorrectly mark an entire conversation as read when reaching the
bottom of a render window, even if there were more messages loaded or
to fetch from the server.
Fix this error in the calculation by asking the correct data
structures if we're actually at the bottom.
To avoid the navigate.js keyboard shortcut code paths circumventing
this new logic, or needing to duplicate it, they now call
process_visible, rather than its helper.
Since we always call `deactivate` from `hashchange`,
`browser_history.state.changing_hash` is always `true` and hence
`save_narrow` just retuns without doing anything.
572443edc6 removed the callsite that triggered the exec in
`zulip::systemd_daemon_reload`, making its inclusion and ordering via
`require` moot.
Remove the call.
The endpoint was lacking validation that the authentication_methods dict
submitted by the user made sense. So e.g. it allowed submitting a
nonsense key like NoSuchBackend or modifying the realm's configured
authentication methods for a backend that's not enabled on the server,
which should not be allowed.
Both were ultimately harmless, because:
1. Submitting NoSuchBackend would luckily just trigger a KeyError inside
the transaction.atomic() block in do_set_realm_authentication_methods
so it would actually roll back the database changes it was trying to
make. So this couldn't actually create some weird
RealmAuthenticationMethod entries.
2. Silently enabling or disabling e.g. GitHub for a realm when GitHub
isn't enabled on the server doesn't really change anything. And this
action is only available to the realm's admins to begin with, so
there's no attack vector here.
test_supported_backends_only_updated wasn't actually testing anything,
because the state it was asserting:
```
self.assertFalse(github_auth_enabled(realm))
self.assertTrue(dev_auth_enabled(realm))
self.assertFalse(password_auth_enabled(realm))
```
matched the desired state submitted to the API...
```
result = self.client_patch(
"/json/realm",
{
"authentication_methods": orjson.dumps(
{"Email": False, "Dev": True, "GitHub": False}
).decode()
},
)
```
so we just replace it with a new test that tests the param validation.
This should give some more room for systems that are still below 4GB
of RAM to use the lower-memory multithreaded mode, which is less
likely to have OOM kills (a very bad experience).
There should be little cost, as few systems are likely allocated with
memory in this range.
So far, there were 2 separate turndown rules for code blocks; one for
general ones, and the other for Zulip message code blocks.
Now the filter rule has been generalised to handle both cases together.
As a side effect, the bug where partially copied Zulip code blocks
lost formatting on pasting has been fixed.
The update_selection function name was rather misleading, since that
function call is in fact what renders the message list object for the
view.
Also add comments about a few subtle/confusing details that I noticed
while debugging this code path today.
As discussed in the new comments, we had a bug where the
system-initiated animated scroll that happens when the compose box
opens as a result of narrowing would race with the internal
rerendering that occurs when the message_fetch request asking the
server for additional data returns.
The correct fix for this is just to open the compose box, if we're
going to do so, before setting the user's scroll position in the
narrowing/rendering process.
This ends up being a UI improvement (in that the compose box is
available for typing a bit earlier) as well as avoiding both the risk
of this race as well as the bad UX of adjusting the user's scroll
position multiple times as part of entering the view.
This does not address an as-yet-unknown bug wherein the animated
scroll that occurs when opening the compose box, when racing with a
background rerender, results in a bogus ending scroll position, though
it's easy to see how that might occur given that rerendering does
clear the DOM briefly.
Needed for typescript, because we want to preserve
types, so instead of mutating a message object,
we can instead calculate return these values
for a message object before it's created in full.
This commit also renames apply_markdown
to render, see this comment
https://github.com/zulip/zulip/pull/28652#discussion_r1470514780
Aman Agrawal
Tim Abbott
Alex Vandiver
Mateusz Mandera
Alya Abbott
N-Shar-ma
Charlie Marsh
Anders Kaseorg
evykassirer