mirror of
https://github.com/fluxerapp/fluxer
synced 2026-10-17 13:05:32 +09:00
fix(app): gate one-Tab message focus behind an experiment (#2762)
This commit is contained in:
1 parent
1281045648
commit
4b278a0da8
35 files changed
+1864
-62
No files matched your search
@@ -5,9 +5,10 @@ use crate::{
|
||||
AppPublicConfigResponse, ExperimentDeliveryConfigResponse, GatewayRolloutConfigResponse,
|
||||
InstanceConfigResponse, InstanceIntegrationsResponse, InstanceMediaResponse,
|
||||
InstancePolicyResponse, InstanceRegistrationResponse, LimitConfigResponse,
|
||||
MessageHoverTrackingConfigResponse, NoiseSuppressionBackend, PendingRegistrationResponse,
|
||||
RegistrationUrlResponse, SsoConfigResponse, VOICE_NS_MAX_GUILD_OVERRIDES,
|
||||
VOICE_NS_MAX_TARGETED_USERS, VoiceNoiseSuppressionConfigResponse,
|
||||
MessageHoverTrackingConfigResponse, MessageKeyboardFocusConfigResponse,
|
||||
NoiseSuppressionBackend, PendingRegistrationResponse, RegistrationUrlResponse,
|
||||
SsoConfigResponse, VOICE_NS_MAX_GUILD_OVERRIDES, VOICE_NS_MAX_TARGETED_USERS,
|
||||
VoiceNoiseSuppressionConfigResponse,
|
||||
},
|
||||
config::AdminConfig,
|
||||
middleware::auth::AuthContext,
|
||||
@@ -149,6 +150,7 @@ pub fn instance_config_page(
|
||||
(gateway_rollout_section(base, csrf_token, &instance_config.gateway_rollout))
|
||||
(voice_noise_suppression_section(base, csrf_token, &instance_config.voice_noise_suppression))
|
||||
(message_hover_tracking_section(base, csrf_token, &instance_config.message_hover_tracking))
|
||||
(message_keyboard_focus_section(base, csrf_token, &instance_config.message_keyboard_focus))
|
||||
(experiment_delivery_section(base, csrf_token, &instance_config.experiment_delivery))
|
||||
@if let Some(limit_config) = limit_config {
|
||||
(limit_config_section(base, limit_config))
|
||||
@@ -1280,6 +1282,110 @@ fn message_hover_tracking_section(
|
||||
)
|
||||
}
|
||||
|
||||
fn message_keyboard_focus_section(
|
||||
base: &str,
|
||||
csrf_token: &str,
|
||||
message_keyboard_focus: &MessageKeyboardFocusConfigResponse,
|
||||
) -> Markup {
|
||||
let status = if message_keyboard_focus.enabled {
|
||||
("Live", BadgeVariant::Success)
|
||||
} else {
|
||||
("Inert", BadgeVariant::Default)
|
||||
};
|
||||
let included_user_ids = message_keyboard_focus.included_user_ids.join("\n");
|
||||
let excluded_user_ids = message_keyboard_focus.excluded_user_ids.join("\n");
|
||||
section_card_with_description(
|
||||
"Message Keyboard Focus",
|
||||
"Picks whether targeted clients run the keyboard navigation rework in the message list. \
|
||||
A targeted client reaches the message list from the composer with one Tab, walks \
|
||||
messages with the arrow keys through revealed blocked groups, and draws the focus ring \
|
||||
inside each row. While the master switch below is off every client keeps the keyboard \
|
||||
navigation it ships with, whatever the rest of these fields say.",
|
||||
html! {
|
||||
form method="post" action={(base) "/instance-config?action=update_message_keyboard_focus"} {
|
||||
(csrf_input(csrf_token))
|
||||
div class="space-y-6" {
|
||||
div class="flex flex-wrap items-center gap-2" {
|
||||
h3 class="text-sm font-semibold text-neutral-900" { "Master switch" }
|
||||
(badge(status.0, status.1))
|
||||
span class="text-xs text-neutral-500" {
|
||||
"Config version " (message_keyboard_focus.config_version)
|
||||
}
|
||||
}
|
||||
(checkbox(
|
||||
"message_keyboard_focus_enabled",
|
||||
"true",
|
||||
"Serve message keyboard focus assignments to clients",
|
||||
message_keyboard_focus.enabled,
|
||||
true,
|
||||
))
|
||||
p class="text-xs text-neutral-500" {
|
||||
"Off is the safe state. With this unchecked every client is told the \
|
||||
rollout is inert and keeps its current keyboard navigation, so the rollout \
|
||||
and targeting fields below have no effect at all."
|
||||
}
|
||||
|
||||
h3 class="text-sm font-semibold text-neutral-900" { "Rollout" }
|
||||
(number_field(
|
||||
"message_keyboard_focus_rollout_basis_points",
|
||||
"Rollout (basis points)",
|
||||
&message_keyboard_focus.rollout_basis_points.to_string(),
|
||||
Some(0), Some(10000), "1",
|
||||
Some("Share of users bucketed into the canary, in basis points: 0 is nobody, 100 is 1%, 10000 is everybody."),
|
||||
))
|
||||
div class="flex flex-col gap-2" {
|
||||
(text_input(
|
||||
"message_keyboard_focus_rollout_salt",
|
||||
"Rollout Salt",
|
||||
&message_keyboard_focus.rollout_salt,
|
||||
"message-keyboard-focus-v1",
|
||||
))
|
||||
p class="text-xs text-neutral-500" {
|
||||
"Seeds the bucketing hash. Changing it reshuffles which users fall \
|
||||
inside the percentage above. Leave it alone to keep the current \
|
||||
cohort stable."
|
||||
}
|
||||
}
|
||||
div class="flex flex-col gap-2" {
|
||||
(textarea_input(
|
||||
"message_keyboard_focus_included_user_ids",
|
||||
"Always-on User IDs",
|
||||
"1500000000000000001\n1500000000000000002",
|
||||
&included_user_ids,
|
||||
4,
|
||||
false,
|
||||
))
|
||||
p class="text-xs text-neutral-500" {
|
||||
"One snowflake per line, or comma separated. These users are targeted \
|
||||
regardless of the percentage above. IDs must contain 1 to 20 decimal \
|
||||
digits. Invalid entries prevent the save; blank entries and duplicate \
|
||||
IDs are ignored."
|
||||
}
|
||||
}
|
||||
div class="flex flex-col gap-2" {
|
||||
(textarea_input(
|
||||
"message_keyboard_focus_excluded_user_ids",
|
||||
"Never-on User IDs",
|
||||
"1500000000000000003\n1500000000000000004",
|
||||
&excluded_user_ids,
|
||||
4,
|
||||
false,
|
||||
))
|
||||
p class="text-xs text-neutral-500" {
|
||||
"Same format. Exclusion wins over both the always-on list and the \
|
||||
percentage, so this is the per-user kill switch."
|
||||
}
|
||||
}
|
||||
|
||||
(form_actions(html! {
|
||||
(submit_button("Save Message Keyboard Focus Configuration"))
|
||||
}))
|
||||
}
|
||||
}
|
||||
},
|
||||
)
|
||||
}
|
||||
|
||||
fn experiment_delivery_section(
|
||||
base: &str,
|
||||
csrf_token: &str,
|
||||
|
||||
Reference in new issue
Block a user