You mentioned failing tests or unexpected values. Can you please clarify? We don't want to test for incorrect values, so if you modified expected test values to make the tests pass, we should talk about it and make sure we use the same expected values as other libraries.

I did not include any tests where the test values are not what they should be (e.g. forcing the tests to past).

However, I don't have any tests for when +lon_0 and +R are being used. I did have tests for the three different rotation methods, in addition to passing +x_0 and +y_0.

Doing some more testing now, seems like +R is fine and it's just lon_0 that is causing problems.

I've added a couple more tests including one that should be failing. That's where the problem is.

I've just modified the code to pass the entire proj4 string to the inner projection, per the original source code. This has not resolved the problem with lon_0. I can see that the inner projection is receiving the correct lon_0 value, and verified that when used on its own (e.g.+proj=moll +lon_0=10 +R=6400000) I can get values I expect when projecting coordinates (so not an implementation issue on the mollweide side, I think).

Neither this code nor the PROJ source code touches lon_0 in either forward or inverse, it's merely passed through as a param to the inner projection, so this is pretty baffling considering I have verified the inner projection is receiving the correct value.

@ahocevar Okay, I have it sorted out now. It seems there is a subtle difference in how PROJ handles lon_0. In the proj4js code here I need to manually apply it before/after forward/inverse.

I also adjusted the param parsing to match the original source (order of preference for which transform method), and added a bunch of tests around param parsing.

Here are the tests that have been added, testing against the same string and coordinates in PROJ:

• forward/inverse for moll, new pole method
• forward/inverse for moll, rotate about point method
• forward/inverse for moll, new equator method
• forward/inverse for moll, new equator method w/ +x_0 and +y_0
• forward/inverse for moll, new equator method w/ +x_0, +y_0, and +R
• forward/inverse for moll, new pole method w/ +lon_0 and +R
• forward/inverse for moll, new pole method w/ +lon_0, +R, +ellps, and +pm
• some of the above but for eqearth
• tests for missing param errors, or incomplete param sets for a given method
• tests for param = NaN errors
• tests for valid inner projection
• tests for invalid param value errors (e.g. +o_lat_1=90)

Let me know any further changes!