Schedule¶

The schedule is a Pretalx-style grid: time runs down the left, rooms run across the top, and each talk is a card whose height is proportional to its duration. Overlapping talks in different rooms sit side by side, so an attendee can see at a glance what is on at any moment and switch days with one click.

Source: talks/views_schedule.py, grid helpers talks/grid_utils.py, template templates/talks/schedule.html. The view is served at /schedule/ and, like the talk list, requires login and scopes everything to the user's accessible events.

How a day is chosen¶

The available days are the distinct dates that have scheduled talks in the user's event scope. Talks left at the FAR_FUTURE sentinel start time (imported but not yet slotted) are excluded by the scheduled() queryset, so they never create a phantom day.

The selected day is resolved in this order:

1. The ?date param, if it is a valid available date.
2. Today, if today has talks.
3. Otherwise the first available day.

A horizontal day picker at the top of the page links each day with a date querystring, and the active day is highlighted.

How the grid is built¶

The grid is plain CSS Grid driven by named row lines. The layout work happens in grid_utils.py and is shared with the session chair grid.

Rows from time boundaries¶

build_grid_slices collects every talk start time and end time for the day into a set of unique boundaries, sorts them, and emits one named grid line per boundary (for example [t-0930]). The gap between consecutive boundaries becomes the row height at 2 px per minute, with a 20 px minimum, so a 45-minute talk is visibly taller than a 15-minute lightning slot. Times are converted to local time before naming, so the labels match the venue clock.

build_time_labels produces the left-column labels (HH:MM) from those same boundaries, skipping the final boundary because it is only an end time with no row beneath it.

Columns from rooms¶

Rooms are collected from the day's talks, de-duplicated, and sorted by name. Column 1 is reserved for the time labels, so the first room lands in column 2. Each talk is then placed with a grid-area of row-start / column / row-end, where the row names come from grid_line_name(start) and grid_line_name(start + duration).

graph LR
    A[Day's talks] --> B[Collect start/end boundaries]
    B --> C["build_grid_slices: named rows + heights"]
    B --> D["build_time_labels: HH:MM labels"]
    A --> E[Collect + sort rooms into columns]
    C --> F[grid-area per talk card]
    E --> F

Streaming prefetch¶

The talks for the day are loaded with with_streamings(), which batch-loads every covering streaming session in one query and caches it on each talk. Without this, the per-card calls to get_video_link(), get_transcription_url(), and has_active_streaming() would each fan out into a separate query per row.

Card rendering and current-talk highlighting¶

Each card links to the talk detail page and shows the title (clamped to two lines), speaker names, a presentation-type badge, and a duration badge in minutes. A bookmark toggle sits in the top-right corner.

The card is tinted by the same timing logic used on the talk list:

• Happening now: orange background with an orange left border (talk.is_current).
• Upcoming: blue background with a blue left border (talk.is_upcoming).
• Past: gray background with a gray left border.

This makes the "now" line of the conference easy to spot while scrolling the grid.

Filtering¶

The schedule has its own filter bar (search box, presentation type, track, and "Saved only"). Unlike the talk list, these filters are applied in Python against the already-built grid items rather than at the database level, because the grid is constructed for the whole day first:

• Search matches the talk title or speaker names (case-insensitive substring).
• Track and presentation type match exactly.
• Saved only keeps talks whose id is in the current user's saved set.

When no filter is active the full day is shown. When filters remove every talk, the page shows "No talks match your filters" rather than the empty-day message, so the attendee knows the day has content they have filtered out.

The event dropdown is always visible and reloads the page with a new event scope; an empty selection falls back to the default event.

Session chair grid¶

Moderators have a parallel grid at /chairs/ that reuses the exact same CSS Grid layout but adds session-chair assignment. This view is moderator-only (staff or superuser); see Questions and answers for who counts as a moderator.

Source: talks/views_chair.py, template templates/talks/chair_grid.html.

What it adds on top of the schedule grid:

• Volunteer or step down: a moderator can claim an unassigned talk as session chair, or release one they already chair. A superuser (admin) can assign any moderator to any talk, or clear it.
• Block highlighting: adjacent talks in the same room with only a short gap between them are tagged with a shared block id. Hovering any talk in a block highlights the whole stretch, and the blocks you chair stay highlighted, so back-to-back sessions read as one assignment.
• Conflict checks: the same person cannot chair two overlapping sessions. An attempt that would overlap is rejected with a message naming the conflicting talks.
• Tight-transition warnings: claiming a talk that starts soon after another you chair in a different room succeeds, but warns that there will be under CHAIR_ROOM_TRANSITION_MINUTES minutes to change rooms.

Assignment posts to toggle_session_chair and, for HTMX requests, swaps the grid table in place with any error or warning attached. The assigned chair's name is only ever shown to moderators, both here and on the talk detail page.