Add an isoline feature. #838
Conversation
manthey
force-pushed
the
isolines
branch
2 times, most recently
from
177e838 to
6f690fd
Compare
|
@manthey is it possible to update this branch? |
|
It still depends on the |
This adds a Math.log10 polyfill and removes the unused Math.sinh polyfill. Add simple isoline example. Add an isoline tutorial. This uses a data set of Oahu's elevation that is lower resolution than the 'dense' set, primarily so that the tutorial will run faster (and test faster). This resolves #592. The `used` function can be overridden for isoline and contour features. This defaults to the `value` function returning a non-null, finite number. Pass the calculated position to the `value` function. This allows the value to be modified based on the position. The isoline label feature is added to the list of dependent features so that something like `myfeature.visible(false).draw()` does what is expected. Don't wrap across longitude if the data is more than 360 units wide, since it probably isn't actually longitude in that case. Reduce memory use for fully populated meshes. For large meshes this can reduce memory use and increase speed. For small meshes, there is practically no difference. Label positions are calculated in map gcs, not interface gcs. Lines are not guaranteed to be linear in the interface gcs, so using it may put labels off of the rendered line. Modify layers for better nested support. Layer's z-index should only be relative to their siblings, not all layers. Layers don't have to be children of the map; they can be children of other layers. Instead of enumerating the map layers for determining siblings, enumerate the children of the layer's parent, filtering to only those that are geo.layer instances. This makes it easier to change the z index of a layer that has dependent layers (such as the isoline feature's canvas layer for text).
| * | ||
| * @returns {this} | ||
| */ | ||
| this.labelPositions = function () { |
There was a problem hiding this comment.
Do we need this as a public function?
There was a problem hiding this comment.
Yes. By default, we recompute local positions after some amount of zooming or panning. If a user wants the labels to be updated on different criteria, they would call this function.
| * labels that need to be shown if there is enough resolution. | ||
| */ | ||
|
|
||
| /* This includes both the marching triangles and marching squares conditions. |
| * `labelViewport` used in generating labels. The object may have no | ||
| * properties if there are no labels. | ||
| */ | ||
| this.lastLabelPositions = function () { |
There was a problem hiding this comment.
Same here, not entirely sure if we need this in the public API
There was a problem hiding this comment.
Yes. As with labelPositions, this information can be used to decide when to regenerate labels if the default is inappropriate.
| } | ||
| } | ||
| result.index.splice(usedPts); | ||
| if (usedPts !== numPts) { | ||
| if (i !== numPts) { |
There was a problem hiding this comment.
Some of the mesh feature was refactored to use less memory by not duplicate arrays if not needed. usedPts is no longer computed while generating a preliminary array, as by skipping that array things are faster. If some vertices are unused, then the previous for loop will have stopped before reaching numPts. In this case we go into the next conditional and exclude the unused vertices.
| * If falsy, they will include the exact minimum and maximum values. | ||
| * @property {number} [spacing] Distance in value units between isolines. | ||
| * Used if specified and `values` is not specified. | ||
| * @property {number[]|geo.isolineFeature.valueEntry[]} [values] An array of |
There was a problem hiding this comment.
Should we mention that either you can use min,max and count or values? I think it is mostly clear, just wondering if we could use bit more strong wording.
There was a problem hiding this comment.
I hope between what is written and the tutorials everything is clear enough. Since values takes precedence, it doesn't mentioned spacing or count, whilst those do mention values.
Prior to this, a custom cmake script (girder.cmake) was used via a direct URL for each external data file. After this change, we use cmake's build-in ExternalData module. This continues to get all external data files listed and automatically untar any .tgz files. External data file references have been moved from testing/test-data to tests/external-data. Eventually I'd like there to be only one of the tests and testing directories, as having two makes things confusing. This also deletes some abandoned cmake scripts and removes some unused data files. As a benefit, data files are now obtained by sha512, which should guarantee that we have the correct file, even if it changes. Before, the download url was used. This makes it so that we only need one entry in the repo per data file instead of two. The only downside is that ExternalData emits a -hash-stamp file in the download directory.
Refactor how we fetch external data.
This adds a
Math.log10polyfill and removes the unusedMath.sinhpolyfill.Add simple isoline example.
Add an isoline tutorial. This uses a data set of Oahu's elevation that is lower resolution than the 'dense' set, primarily so that the tutorial will run faster (and test faster).
This resolves #592.
This depends on a variety of open PRs (and includes them as part of this PR) and could be rebased once those are merged it.