This file is automatically exported from /organizers-notebook/index.org. You might prefer to navigate this as an Org file instead. To do so, clone the wiki repository.

This document is the general organizers' notebook that can be used as a starting point each year. For specific years, see:

Table of Contents

Starting up the conference planning process

Check in with core organizers from the previous year to see who’s in.

Create the private repository

This repository will store submissions, contact information, etc.

  1. Create the repository on front0. Ex:

    sudo su - git
cd repositories
mkdir emacsconf-2024-private.git
cd emacsconf-2024-private.git
git init --bare

  2. Edit gitolite-admin/config/gitolite-conf. Add lines for the group and the repo. Commit and push. Ex:

    @emacsconf-2024-org = bandali sachac ...
repo emacsconf-2024-private
     RW+ = @admins @emacsconf-2024-org orga
  3. Clone the private repository into the appropriate location. git clone git@git.emacsconf.org:emacsconf-2024-private private

Create the public organizers’ notebook in the wiki

Ex: 2024/organizers-notebook/index.org

Phases and lessons learned

Draft CFP

Create linked pages

  • ☐ year index
  • ☐ volunteer page
  • ☐ submit page

How to mark pages as drafts

Put inside double square brackets: !template id=pagedraft

Previous years

2023

Lessons learned

  • Maybe incentivize proper timezone specification by saying we can translate times to their local time?
  • Make sure to include cfp.org as an attachment instead of inline
  • bandali likes having the commitment to freedom section in the CFP as a form of activism.
  • Ask for public e-mail or contact information, IRC handle in CFP.
  • Nudge everyone towards 5/10 or 20 minutes, and consider < or > on a case-by-case basis.
    • In previous years, we offered 40 minutes as a regular choice, but that was difficult because we had to figure out how to squeeze lots of talks into the schedule and we didn’t know what could be shortened. In 2022, we asked people to provide quick outlines for 10-, 20-, and 40-minute formats if they wanted a long talk slot, which was helpful in prioritizing. In 2023, we just suggested 5/10 and 20-minute slots knowing that there would probably be plenty of time for Q&A, with some longer talks approved on a case-by-case basis. Because talks are (ideally) prerecorded, 20 minutes seems to be a good length to get into something interesting while leaving time for Q&A so that people could dive deeper into particular things that interested them.
  • Rolling acceptances were very nice. We gave people a week to respond, which felt a little quick sometimes, but gave people enough time to speak up to see if they had any red flags or suggestions.
  • We didn’t extend the CFP period in 2023, great!
  • Two streams worked out well. Sunday morning had network congestion issues, so keep an eye out for those.
  • It’s easier for us to extend beyond 5pm than to go before 9am (especially for Pacific time). Extending beyond 5pm puts strain on European organizers and volunteers, though.
  • Put the speaker introduction right after the talk description for easier web page copying.

Announce

  • Post CFP to emacsconf-discuss, reddit.com/r/emacs, and in Emacs News
  • Submit to https://foss.events/

Process proposals

  1. In the private repository, create a conf.org.
  2. Set emacsconf-org-file to its location.
  3. Create a heading for Submissions and set CUSTOM_ID to submissions.
  4. Prepare the prepare and submit pages on the wiki.

When proposals come in:

  1. Acknowledge with an e-mail along the lines of:

    Hello, ....!

<feedback> We'll wait another week in case anyone else wants to
chime in before I add it to the wiki and send further notes. :) Thanks for
submitting this!

  2. Use emacsconf-mail-add-submission if it works, or create the entry and paste it in. org-toggle-item can help convert Org-style headings into a nested list.

  3. Set the following fields (my/org-set-property can be helpful): EMERGENCY, Q_AND_A, AVAILABILITY, NAME, PRONOUNS, TIME, SLUG, EMAIL, NAME_SHORT, CUSTOM_ID, TRACK, TIMEZONE, CATEGORY, DATE_SUBMITTED, PUBLIC_EMAIL

Accept proposals

Setup:

  1. Create YYYY/talks.md.

For each talk:

  1. Set up “Talk abstract” and “Talk details” sections in the conf.org talk subtree.
  2. Change status to TO_CONFIRM.
  3. Create wiki page with emacsconf-publish-add-talk.
  4. Use emacsconf-publish-info-pages and emacsconf-publish-schedule as needed.
  5. Send acceptance e-mail with emacsconf-mail-accept-talk

Draft schedule

  • Create entries for opening and closing remarks
    • FIXED: t
    • SCHEDULED
    • TIME: 10
    • BUFFER: 0
    • CUSTOM_ID:
    • SLUG:
    • Q_AND_A: pad
  • Create draft-schedule section in the public organizers notebook for the year
  • Add missing items to plan

Set up backstage and upload

See README.org.

Lessons learned:

  • Next year, let’s get upload.emacsconf.org in place as the domain name so that people don’t try to FTP files to it.

Post the schedule

  • <elisp:(setq emacsconf-publishing-phase 'schedule)>
  • <elisp:emacsconf-publish-info-pages>
  • <elisp:emacsconf-publish-schedule-org-files>
  • <elisp:emacsconf-ical-generate-all>
  • Update talks.md to refer to watch.md as well

Process uploaded files

Basic setup:

  1. Set up the media server and symlinks: README.org (Media).
  2. Set up the prerec processing areas on res: README.org (Processing prerecs).
  3. Set up the caption processor on res: README.org (Captioning).

Process:

  1. Review uploaded files with <elisp:emacsconf-upload-dired>
  2. Open the JSON for the file you want to process and use <elisp:emacsconf-upload-copy-from-json> . If more than one file uses the same extension, provide a suffix to disambiguate them.
  3. On res:
    1. Create a directory ~/current/files/$slug.
    2. Copy the uploaded file from ~/cache as $video_slug--original.$extension, or use rename-original.sh $slug $file.
    3. Call process-prerec.sh $file.
  4. When the screens are finished, do locally:
    1. update-cache on my computer.
    2. <elisp:emacsconf-publish-cache-video-data> or <elisp:emacsconf-cache-all-video-data>
    3. <elisp:emacsconf-publish-backstage-index>

Generate assets

In-between pages

  • Make the $year/assets/in-between directory and copy the template.svg from the previous year.
  • <elisp:emacsconf-stream-generate-in-between-pages>

Intros

  • In an empty VTT file, <elisp:emacsconf-subed-intro-subtitles>
  • Record the audio and get captions
  • Splice together into a file that will be processed with subed-record-compile-video.

Prepare the infrastructure

Checklist:

  • ☐ IRC
    • ☐ Manual announcements
    • ☐ Automated announcements
  • ☐ Web conference
  • ☐ Streaming assets
    • ☐ Overlays
    • ☐ Intros
  • ☐ OBS
  • ☐ VNC
  • ☐ Publishing to the wiki
  • ☐ Publishing to the media server
  • ☐ Etherpad
  • ☐ Mumble
  • ☐ Icecast

Getting ready for captions

Onboarding:

  • Add captioner to conf.org with the :caption: tag, under a Volunteers heading with the volunteer tag. Make sure the EMAIL property is set.
  • Add captioner to emacsconf-org mailing list.
  • Send backstage information <elisp:emacsconf-mail-backstage-info-to-volunteer> . (Can be sent to all captioning volunteers with <elisp:emacsconf-mail-backstage-info-to-captioning-volunteers>)

Getting ready for the conference

<elisp:emacsconf-pad-prepopulate-shift-hyperlists>

TODO Send check-in details, schedule update, etc.

Two days before the conference

Go through the :tminustwo: tags.

One day before the conference

Go through the :tminusone: tags.

During the conference

<elisp:emacsconf-pad-open-shift-hyperlist>

General infrastructure

Icecast

Success: You can use OBS+VNC to record, and the results can be viewed by mpv.

  • ☑ Gen
  • ☑ Dev

Also, recordings are available in /data.

New year:

  1. Reprovision with

    ansible-playbook -i inventory.yml prod-playbook.yml --tags stream

    so that the year is updated in the configuration.

  2. SSH to live0.emacsconf.org and run /etc/init.d/emacsconf stop. Confirm that icecast has stopped with ps | grep emacsconf; if not, kill the process. Start it again with /etc/init.d/emacsconf start

  3. To test, you can use FFmpeg like this for the gen stream:

    ffmpeg -y -f lavfi -i testsrc=size=1280x720:rate=10 -i background-music.opus \
                         -c:v libvpx -b:v 500K -crf 25  -b:a 128k \
                         -content_type video/webm "icecast://emacsconf:$EMACSCONF_ICECAST_SOURCE_PASSWORD@live0.emacsconf.org:8001/gen.webm"

    For dev:

    ffmpeg -y -f lavfi -i testsrc=size=1280x720:rate=10 -i background-music.opus \
                         -c:v libvpx -b:v 500K -crf 25  -b:a 128k \
                         -content_type video/webm "icecast://emacsconf:$EMACSCONF_ICECAST_SOURCE_PASSWORD@live0.emacsconf.org:8001/dev.webm"

    and then view it with

    mpv https://live0.emacsconf.org/gen.webm

Watching pages

Success: You can watch at https://live.emacsconf.org.

  • ☑ https://live.emacsconf.org goes to the current year’s page.
  • ☑ gen stream works in the browser
  • ☑ dev stream works in the browser
  • ☑ Schedule is correct

New year:

  1. ansible-playbook -i inventory.yml prod-playbook.yml –tags live
  2. Republish with <elisp:emacsconf-publish-watch-pages>.
  3. Update front:/var/www/live.emacsconf.org.

Set up test streams (see ffmpeg instructions from Icecast) and then:

VNC

We use VNC to connect to the X servers on res.emacsconf.org so that we can stream from it.

Success:

  • ☑ Confirm that you can connect to emacsconf-gen via VNC
  • ☑ Confirm that you can connect to emacsconf-dev via VNC

Setting up

  • <elisp:emacsconf-publish-res-index>

Instructions

NOTE: VNC+OBS doesn’t work well if you have a window manager that automatically resizes windows, like i3. Please configure your window manager so that the VNC window is not resized.

  1. Install a VNC viewer on your system (ex: tigervnc-viewer).

  2. Set up your local environment:

    • gen: export TRACK=gen; export TRACK_PORT=5905; export SSH_PORT=46668
    • dev: export TRACK=dev; export TRACK_PORT=5906; export SSH_PORT=46668

  3. Copy the password:

    scp emacsconf-$TRACK@res.emacsconf.org:~/.vnc/passwd vnc-passwd-$TRACK -p $SSH_PORT

  4. Forward your local ports and connect via VNC viewer to the appropriate forwarded port from your laptop:

    ssh emacsconf-$TRACK@res.emacsconf.org -N -L $TRACK_PORT:127.0.0.1:$TRACK_PORT -p $SSH_PORT &
sleep 5   # Give it time to establish the tunnels
xvncviewer 127.0.0.1:$TRACK_PORT -shared -geometry 1280x720 -passwd vnc-passwd-$TRACK &

If you get the following error:

channel 2: open failed: connect failed: Connection refused
CConn:       End of stream
CConn:       The connection was dropped by the server before the session could
             be established.

then the VNC server hasn’t started yet. You can start it with

ssh emacsconf-$TRACK@res.emacsconf.org -p $SSH_PORT /home/emacsconf-$TRACK/bin/track-vnc

and then connect with:

xvncviewer 127.0.0.1:$TRACK_PORT -shared -geometry 1280x720 -passwd vnc-passwd-$TRACK &

Streaming with OBS

Success: Confirm that you can stream

  • ☑ gen
  • ☑ dev

New year: reprovision with

ansible-playbook -i inventory.yml prod-playbook.yml --tags obs

so that the year is updated in the shell scripts.

Instructions

  1. Connect to the VNC session for the track.

  2. Start recording (not streaming). If you don’t see OBS when you connect, it’s probably on workspace 2, so you can switch with Alt-2. If you still don’t see it there, you can open a terminal with Alt-Enter and then run track-obs. After you start recording, confirm that it is now broadcasting to the stream.

  3. Verify with MPV on your local system:

    mpv https://live0.emacsconf.org/$TRACK.webm &

Icecast

Success: You can use OBS+VNC to record, and the results can be viewed by mpv.

  • ☐ gen
  • ☐ dev

New year: reprovision with

ansible-playbook -i inventory.yml prod-playbook.yml --tags stream

so that the year is updated in the configuration.

This is on live.emacsconf.org and can be restarted with /etc/init.d/emacsconf restart.

Shell scripts

Success: From the commandline, hosts can:

  • ☑ play just the intro: intro $SLUG
  • ☑ play a file without an intro: play $SLUG
  • ☑ play a file with an intro: play-with-intro $SLUG (hmm, this didn’t show the subtitles) ah, it’s because the intros need to be in the cache dir?
  • ☑ open the Q&A for a talk: handle-qa $SLUG
    • ☑ open the pad: pad $SLUG
    • ☑ open BBB: bbb $SLUG
  • ☐ do the whole thing: handle-session $SLUG
  • ☐ rebroadcast the other stream:
    • rebroadcast gen
    • rebroadcast dev
  • ☐ control background music: music, stop-music

Setup:

  • OBS must be streaming from the stream you’re testing. Streaming with OBS
  • You may want to generate test assets and copy them over to the assets/stream directory.
  • Generate all the overlays and copy them to assets/overlays
  • Set up the assets/music directory

Backstage

Update this with <elisp:emacsconf-publish-backstage-index>

Explanation of files:

–original.mp4/mov/webm original file as uploaded by the speaker
–silence.mp4/mov/webm silence sample if uploaded by speaker
–reencoded.webm converted to .webm format and compressed
–reencoded.opus audio only
–normalized.opus normalized audio
–reencoded.vtt WhisperX captions, WebVTT format
–reencoded.json WhisperX word data
–reencoded.srt WhisperX captions, SRT format
–reencoded.txt WhisperX captions, plain text
–main.vtt edited captions
–main–chapters.vtt chapter markers
–main.txt captions as text (optional)
–main.webm version for streaming based on the most recent resources

Other files might also have been uploaded by the speaker, such as slides or notes.

IRC web client

Success:

Remember to schedule a task to ask libera.chat to increase the number of users it will accept from chat.emacsconf.org to avoid connection exhaustion errors.

Message I’ve posted to support@libera.chat:

Thank you for running libera.chat. We’re looking forward to using IRC (#emacsconf-gen, #emacsconf-dev) for the discussions for the upcoming EmacsConf (Dec 7-8, https://emacsconf.org). In the past, we’ve needed to coordinate with Libera staff to increase the number of connections allowed from chat.emacsconf.org during conference days in order to avoid connection exhaustion errors. Is that something we can plan now or would you like me to bring it up a few days before the conference?

Etherpad

Success:

  • ☑ pad.emacsconf.org redirects to this year’s entry
  • ☑ Each talk has its own Etherpad

Publishing media to the server and to the wiki

Start of year:

  1. Set media_protect_root to true in Ansible group_vars/all.yml.
  2. ansible-playbook -i inventory.yml prod-playbook.yml --tags media
  3. Generate the index with emacsconf-publish-update-media

Confirm by setting a submitted talk to PLAYING and testing with <elisp:emacsconf-publish-media-files-on-change> . The public media directory should have the files and the entry should be in the index. Switching it back to TO_STREAM and calling <elisp:emacsconf-publish-media-files-on-change> should remove it.

TODO Switch public media to unprotected root before the conference :tminustwo:

  1. Clear public media directory.
  2. Set media_protect_root to false in Ansible group_vars/all.yml.
  3. ansible-playbook -i inventory.yml prod-playbook.yml --tags media

You can generate the index with emacsconf-publish-update-media.

Automated IRC announcements

Success:

  • ☐ When a talk starts playing, it is announced in the relevant channel

BigBlueButton

ssh orga@media.emacsconf.org “~/bin/bbb-before sat-open” https://media.emacsconf.org/2024/current/bbb-sat-open.html

ssh orga@media.emacsconf.org “~/bin/bbb-open sat-open”

After restarting, and in case of “Firefox can’t establish a connection to the server at wss://bbb.emacsverse.org/graphql.”

ssh root@bbb cd greenlight-v3 docker compose restart

Does not seem to work in Firefox private browsing; documented at https://emacsconf.org/2024/watch/.

System audio limitations

Sharing system audio by changing the audio device to “Built-in Audio Analog Stereo” is very choppy, only gets captured in the recording if I’m using Firefox (not Chrome) AND there’s at least one listener, and doesn’t allow simultaneous narration from the speaker.

Using pulseaudio to change the app’s recording device to the audio monitor results in smoother audio (because it’s pretending to be the microphone), but still doesn’t allow simultaneous narration from the speaker.

Combining microphone + system audio into a new virtual microphone source is probably more complicated than we want to walk speakers through.

BBB says sharing system audio works only if you use Chrome on Mac OS X or Microsoft Windows (https://support.bigbluebutton.org/hc/en-us/articles/1500005315642-Can-I-play-a-video-using-screen-sharing). I tested it on Linux and sharing system audio is not an option. If you happen to know of any more elegant way to do this, I’d love to hear it.

Related: https://github.com/bigbluebutton/bigbluebutton/issues/8632

Manual IRC announcements

Success:

  • ☑ You can /opall, /conftopic, and /broadcast

Low-res stream

Should be automatic, handled by Icecast backup on live:

  • /usr/local/bin/emacsconf-lowres-gen-on-connect
  • /usr/local/bin/emacsconf-lowres-dev-on-connect

Upload talks to YouTube

Channel videos

(seq-keep
 (lambda (o)
     (when (and (string= (plist-get o :status) "TO_STREAM")
                            (not (plist-get o :youtube)))
         (list
            (plist-get o :slug)
            (org-link-make-string
             (concat "copy:"
                             )
             "video file")
            (org-link-make-string
             (format "elisp:(emacsconf-publish-video-description \"%s\" t)"
                             (plist-get o :slug))
             "desc")
            (if (emacsconf-talk-file o "--main.vtt")
                    (org-link-make-string
                     (concat "copy:"
                                     (emacsconf-talk-file o "--main.vtt"))
                     "sub file")
                "")
            (org-link-make-string
             (concat "elisp:(emacsconf-set-property-from-slug \""
                             (plist-get o :slug)
                             "\" \"YOUTUBE\" (read-string \"YouTube URL: \"))"
                             )
             "save")
            (plist-get o :scheduled))))
 (emacsconf-publish-prepare-for-display (emacsconf-get-talk-info)))

Restream to YouTube

Other process notes

Adding another user to BBB

ssh bbb.emacsverse.org
docker exec -it greenlight-v2 bundle exec rake user:create["USERNAME","EMAIL","PASSWORD","user"]

When a talk is added after the schedule has already been drafted

  • ☑ Add the talk to conf.org
  • ☑ Add the talk ID to the draft schedule
  • ☑ Run the draft schedule block
  • ☑ <elisp:emacsconf-schedule-update-from-info>
  • ☑ <elisp:emacsconf-publish-talks-json-to-files>
  • ☑ Generate pads with ansible-playbook -i inventory.yml prod-playbook.yml --tags pad
  • ☑ <elisp:emacsconf-set-file-prefixes>
  • ☑ <elisp:emacsconf-update-schedule>
  • ☑ <elisp:emacsconf-stream-generate-overlays>
  • ☑ <elisp:emacsconf-stream-generate-test-videos>
  • ☑ rsync -avze ssh ~/proj/emacsconf/2024/assets/ res:/data/emacsconf/shared/2024/assets/
  • ☑ Send the upload instructions
  • ☑ Generate the intro text
  • ☐ Record the intro

When a talk is cancelled

  1. Update the status to CANCELLED.
  2. Rejig the schedule if needed, and update with <elisp:emacsconf-schedule-update-from-info>
  3. Update the published schedule with <elisp:emacsconf-update-schedule> and push the wiki.
  4. Regenerate the intros (before, after). <elisp:emacsconf-stream-generate-in-between-pages>, then use <elisp:subed-record-compile-video> in the intros file.
  5. Sync the new intros to the /data/emacsconf/shared/$YEAR/assets/intros directory.
  6. Regenerate the pads: <elisp:emacsconf-pad-prepopulate-all-talks>

When a talk Q&A changes method

  1. Update the talk property.
  2. Update the schedule with <elisp:emacsconf-update-schedule>
  3. Regenerate the intros (before, after). <elisp:emacsconf-stream-generate-in-between-pages>, then use <elisp:subed-record-compile-video> in the intros file.
  4. Sync the new intros to the /data/emacsconf/shared/$YEAR/assets/intros directory.
  5. Regenerate the pads: <elisp:emacsconf-pad-prepopulate-all-talks>
  6. Regenerate talks.json: <elisp:emacsconf-publish-talks-json-to-files>

When a system is down

Let people know in #emacsconf and #emacsconf-org

Update the status page:

<file://ssh:orga@front0.emacsconf.org:/var/www/status.emacsconf.org/index.html>

/broadcast may also be helpful

When we need to bring on a last-minute volunteer

/backstage nick

will give them the backstage credentials and links to the index

Checking people in

/checkin nick /room nick (should be automatic if live, but there just in case)

Ansible

git clone git@git.emacsconf.org:pub/emacsconf-ansible