Skip to content

Conversation

ear7h
Copy link
Contributor

@ear7h ear7h commented Jun 29, 2019

The debugger code for this repository (internal) was updated with what was in gdey/quad-edge and the most of the previous triangulation code from this repo was carved out and and replaced by gdey/quad-edge. (most bc the other repo had some dependencies on this one).

The refactor was mostly done by a copy and a simple search and replace for import paths.

gdey and others added 19 commits September 14, 2018 15:53
The expected values of the tests were wrong. Fixes to the wkt package
did not fix the tests.
The constraineddelaunay package had some bugs that were diffcult
to track down as it is trying to handle the general case. For the
Hitmap based makevalid we don't have to handle the general case.
We can assume that all segments will not intersect except at their
end points.

The original routines returned basic geom types, instead of Line
and Triangles; which then had to be converted back and forth. This
one now takes a set of points and segments (instead of a geometry) and returns
a set of Triangles.

Added test cases and helper methods.
When we are ready to work on the setdiff and a generic
constraineddelaunay triangulation we can revert this commit.

This is more to make sure that we have good test that are in a
passing state. Currently several of tests for triangulate and
makevalid/setdiff fail, as that code is still being developed.

From this package it turned out that only package that was needed
was the the quadedge package. So, this commit removed the other
packages that were brought in from the merge.
1. Added doc on how to add test cases.
2. Added and updated test functions to new style
Segments were not getting generated correctly.

1. Broke out the code to filter segments using the clipbox
2. Broke out the code to split and then filter the segements using
clipbox
    2a.  Stop adding the endpoint of the lines and only use the
    the intersecting points. Not sure if this is going to create
    an issue for others.
    2b. Not trying to filter out and split the segments at the
    same time. Maybe a little less efficent

[issues] Issues still remaining

Looks like the hit map for the test case is still wrong. So,
the triangles that are generated seem correct, but the hit map
is incorrectly identifying the triangle.

Need to fill in the correctly expected values
* [testing] added some testcases incl. images

* [testing] re-ordered testcases

* [testing] some more testcases and images
Added a new package that allows one to capture geometeries through
out a process into a spatialite database. This can then be used
to visulaize the process or test case.
Test broke with a few changes dues to precesion.
Plumbbed debuggere deeper into the code.
* added description for testing in docker-env

* added new test-case (9)

* updated docker-info

* added QGIS styles

* turn off debugging
@ear7h ear7h changed the base branch from gdey_delaunay to master July 3, 2019 23:44
@gdey
Copy link
Member

gdey commented Jul 10, 2019

This looks pretty good. I fixed the conflict on this branch: (https://github.com/go-spatial/geom/tree/ear7h-gdey_delaunay) but I started to look into geopackage (https://www.geopackage.org/spec121/index.html) as a replacement for spatialite. If you (@ear7h) want to take a look at it, that would be good.

@gdey
Copy link
Member

gdey commented Aug 11, 2019

With port_constrained branch all parts (except some make valid tests) of this have been split out and recombined into bite-size parts. Closing.

#49
#50
#51
#53
#54

@gdey gdey closed this Aug 11, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants