• “Story page” - show images, audio, wait for response
• “Video page” - show video (possibly also images? combine?)
• Auto-advance option
• Replay option
• When advance is allowed
• Option of frame with parent text
• Show/toggle webcam?
• Allow choice of image/option, entry of text, or completion of form required to move on

• Allow specification of whether to do recording throughout preview
• Improve formatting
• Progress bar

Allow specification of frame where video recording should start and stop, across frames. Will need to handle video recording being started/stopped in an intermediate frame, never being explicitly stopped, etc.

• Allow use of substitution in frame without randomizing
• Ensure that a 2x2x2 factor design is easy to set up
• Allow researchers to set up a simple reusable "frame group" that could be used multiple times in a study, with parameters overridden if needed

Allow more flexible and compact specification of stimulus locations

line 787, exp-player/components/exp-lookit-video/component.js - e.$(...) is undefined
Not consistent but have observed multiple times, and see on Sentry on prod
Probably due to the interval timer not being removed correctly in all cases when moving to next frame

• Additional testing of F1 option
• Provide alternate route to leaving early for various keyboards

• Allow debriefing to use condition information
• Allow debriefing to use session data more generally

Esp. for errors researchers can cause: e.g., incorrect inputs to a frame.

Possibly provide w/i experimenter interface (i.e. as analogues to exp-player component):

• parse JSON / generate simulated sequence of frames
• check for resources (images/video)

• "Cycling through" randomizer (fixed or randomly assigned starting point, specify what counts as completed; e.g., all kids do sessions 1, 2, and 3)
• Shuffle across sessions (e.g. all kids do sessions A, B, and C but in random order)
• "Always assign to same condition" randomizer

Not sure on implementation details yet, but some way for researchers to provide and arrange multiple elements from a reasonable list (e.g., a video plus a survey! their own webcam view plus some instructions, and optional record buttons!) would allow substantially more flexibility without having to create new frames. Potentially provide set of elements and their positions.

Dynamic font sizing, e.g. http://jquery-textfill.github.io/ or https://github.com/BrOrlandi/big-text.js. See https://stackoverflow.com/questions/687998/auto-size-dynamic-text-to-fill-fixed-size-container

Currently hangs due to phantomjs issue

Allow user-supplied function that maps from existing frame data onto parameters for this frame. May at same time create simple randomizer that takes a list of frames and an index so that one of the things the user-supplied function can do is determine which frame to go to (e.g., depending on eligibility as determined in survey frame, or depending on whether we've detected N correct answers in a row, etc.). Note that this will allow keeping track of state since parameter generated for this frame can then be used next time.

Informative error messages upon trying to parse randomizer frame. Include e.g. parameterSetWeights length validation.

Right now only data from the last time the frame was viewed ends up saved, and the sequence recorded doesn't clearly indicate the order.

• Allow static image display w/ some easy default for central, left/right
• Cover alternation designs
• Cover video display
• Implement general “segments” (e.g. calibration, attention-getter)
• Allow more general specification of pause behavior
• Allow displaying or toggling webcam on pause
• Specify fullscreen behavior
• Specify when to require recording

Finish functionality, clean up as needed, generalize

Test more thoroughly and generalize

• Test current functionality and clean up as needed
• Create assent variant
• Allow using consent-only or consent + assent depending on age

• Test current functionality and fix issues as needed
• Create a small library of standard surveys and publish

Allow randomization with a goal of maintaining a given ratio of condition assignments

• User specifies what counts as complete data record
• User specifies whether to count unique participants only
• Allow matching on specified child criteria (e.g., age, sex)

Figure out a more sensible and maintainable workaround for frames that lead to fullscreen vs. non-fullscreen following frames; right now frames that precede FS frames have to be of type unsafe.

Decide whether it makes sense to switch to ember-cli-addon-docs for frame documentation in exp-addons. The current docs technically include all necessary information but it's somewhat hard to interpret. (If so, will add additional "actually switch" task.)

Notes: See https://medium.com/build-addepar/ember-cli-addon-docs-shared-documentation-for-the-ember-ecosystem-6f29aa0cee87. We would still be using YUIdoc syntax for frame documentation, but my impression is that the categories we could keep track of might fit more naturally with the components we're writing, rather than having to e.g. use the serializeContent method to describe what info each frame sends to the server. (Or it might be just as much of a mess! But seems worth looking into.)

Current frame types:
exp-video-config (ok)
exp-video-consent (-> exp-lookit-video-consent)
exp-physics-intro (-> exp-lookit-text)
exp-exit-survey (-> exp-lookit-exit-survey)
exp-mood-questionnaire (-> exp-lookit-mood-questionnaire)
exp-video-preview (ok)
exp-video-config-quality (ok)
exp-physics-pre-video (-> exp-lookit-instructions)
exp-physics-preview-explanation (-> exp-lookit-preview-explanation)
exp-video-physics (-> exp-lookit-video)

Safely allow entry of HTML/CSS.

Right now frames are named using position in the entire sequence + highest-level frame name, e.g. 32-test-trials; would like to allow use of sub-names within randomizers.

Create a frame that shows previous sessions of the same stud with the same child, along with any feedback