Check out e.g. http://qunitjs.com and https://github.com/jonkemp/qunit-phantomjs-runner

The hack in timelapse.js to slow down Mac scrolling is useful for laptop scrollpad and "magic mouse" with scrollpad on the mouse, but is way too slow for Macs with mechanical wheels. This isn't a problem with zooming UIs on similar platforms like Google Maps.

Should we upgrade to a more modern interpretation of WheelEvent honoring DOM_DELTA_PIXEL vs DOM_DELTA_LINE?

Make sure to rebase so that this repo doesn't remain 290Mb...

Fake data format that serves random data.
Also: Generalize TiledBinFormat to handle any format class for tiles.

It happens with two different vector layers

https://earthtime.org/stories/allegheny_voting_2018#waypointIdx=4 should show red and blue dots, but shows white instead. ("points" binary file)

https://earthtime.org/stories/pittsburgh shows green dots correctly, but blue turns white ("dotmap" style layer). Or later, own v rent, all dots are white.

Landsat not loading properly underneath seems like a regression related to a bug fixed by artificially introducing "pause" to the landsat layer.

Suspicious of the video sync merge from headless.

"Starting time" and "Ending time" labels don't make sense in the dialog.
Selecting start 0 and end 4, video fails to generate (zero frames)

connect time zoom to inverse spatial zoom - when zoomed out, have small time tiles, when zoomed in, send all/nearly all time in one chunk.

Right now we only simplify to each region becoming a triangle. Need to merge close-by regions together on extreme zoom out.

I'd like us to start the big cleanup by making a list of visualizations, and explaining what they are (for), and then deleting any that are old and not really relevant any more. here are the files that I'd like comments on (you can just make the list in this ticket):

• ./viirs/00.html
• ./viirs/02.html
• ./viirs/06.html
• ./viirs/01.html
• ./viirs/03.html
• ./viirs/05.html
• ./viz.html
• ./frac-focus.html
• ./google-maps.html
• ./index.html - this is the current version used by skytruth, generalized to handle any size of dataset, no hard-coded filenames or the like

Parent and child tiles are sometimes drawn at the same time during zoom, when a parent is loaded and fewer than all 4 children are loaded. The parent and partial children are all drawn. For raster without transparency, the child covers the parent, for a good result. But with vectors or alpha, sometimes lines don't line up (e.g. WDPA) or things become brighter (e.g. dotmaps). Ideally we should either not draw a partial set of children when the parent is present, or draw partial parent where children don't overlap . The latter might be unworkable.

Allow for csv data to sometimes specify lat/lon to but still do country centers for everything else. Right now if you specify the use of lat/lon, you have to do it for all entries.

Not sure if this is related to #44, but when putting a period in the category name (i.e. Andrew Sandbox -> Andrew Sandbox 2.0) as in my spreadsheet: https://docs.google.com/spreadsheets/d/1UXgnuncXD27FcXCPgDu9DMh29elzkDUq0_CvyixAstY/edit#gid=870361385
the following issue happens:

Similarly, if I add an apostrophe (i.e. Andrew's Sandbox), all the data layers disappear.

Add an enhanced time-domain control that works and looks more like the time slider in the timelapse tool with a graduated time slider and a play/pause control.

Spinner management code, such as WebGLVectorTile2._removeLoadingSpinner and WebglMapTile._removeLoadingSpinner, slows image tile loading by factor of three, due to the complex dom search. It would be better to manage the spinner without use of the dom.

For large data sets, it is desirable to enable display of some data while the rest of the data is still loading. Add support for processing incoming data in chunks instead of waiting for the entire data set to be downloaded.

Proposal: implement a simple form of inclusion -- optional rows in a spreadsheet that could look something like "include https://docs.google.com/spreadsheets/d/1rCiksJv4aXi1usI0_9zdl4v5vuOfiHgMRidiDPt1WfE/edit#gid=870361385".

The idea is that if you are developing new non-public CSV or dotmap additions (e.g. davos2019), you could put only the additions and changes in your spreadsheet tab, then include the corresponding main public tab.

...

I like the simple form of inclusion idea. It seems to minimize the number of steps and reduces the chance at human error that was possible when manually copying and pasting from different spreadsheets. I’m all for it!

UI to change color/magnitude dynamically based on data fields (score)

Involves mirroring layers, displaying in webgl

"Randy's story" in this spreadsheet
https://docs.google.com/spreadsheets/d/1AaJtLP9Xzr4EHMNfFmBaYeqUQ3IVHjfnC_3ruaTjfLs/edit#gid=0

works for desktop:
https://earthtime.org/stories/randys_story#waypoints=1AaJtLP9Xzr4EHMNfFmBaYeqUQ3IVHjfnC_3ruaTjfLs.0

but not mobile:
https://earthtime.org/m/stories/randys_story#waypoints=1AaJtLP9Xzr4EHMNfFmBaYeqUQ3IVHjfnC_3ruaTjfLs.0

In this spreadsheet both the theme and the story have apostrophes -- not sure if that's significant.

By removing both apostrophes, the mobile started working.

Download took 2.5 mins at 3 mbps (SFO airport wifi)

When the user selects a waypoint by clicking or fwd/back, use that waypoint for the auto mode, instead of keeping an independent auto mode state. So if the user clicks on the 10th waypoint and pressed L, it should move to the 11th.

Consider the case where the markers on the map represent physical features that are 1000m across. In that case, when you zoom in very far, the marker should become larger in pixel size once you reach the point where the default display size becomes smaller than the physical size in the map projection.

This could be accomplished by supplying a Minimum Size attribute that is used to compute the display size of markers. This attribute would be expressed in map projection units.

Maybe have a "loading" spinner next to the layer legend?

VIIRS is an example use case; it's relatively large and can take over a minute to load over a slow network connection.

Currently the video free-runs and is shown starting at a random time when the user scrolls to it.

From Mark:

I've written a brief description of the dataset and included it in the README.md file, which you can view at https://github.com/CMU-CREATE-Lab/drilling-data-tools

As before, you can explore the data here: https://cmu-create-lab.github.io/drilling-data-tools/national-database/wells.html

And download the .bin file here: https://github.com/CMU-CREATE-Lab/drilling-data-tools/blob/master/national-database/data/data.bin

json-to-bin.py started out as a straight port, but has now diverged from json-to-bin.rb as it implements the new format with a json header, and json-to-bin.rb has been updated with new changes from another repo to support more options.

Could we somehow merge this? Esp, could we settle on one binary format?

Consider serving tiles in binary or if parsing is needed and expensive, consider web workers.