Update to D3 4.0. about d3-geo-projection HOT 22 CLOSED

I’ve decided that rather than try to apply some variable padding to each projection, to just fit them exactly into 960×500 using the new test/compute-scale.

from d3-geo-projection.

• d3.geoBottomley done locally

[UPDATE: SOLVED] Before I continue I have an issue with the image resolution for the tests: the test output is visually OK but not the same as those in img/

For reference I'm attaching the result of npm test for mollweide: it's 209KB instead of 82.6KB for img/mollweide.png

from d3-geo-projection.

instructions from @mbostock to play this game:

Take a look at the past commits to get a sense of what’s needed.
e.g., 2ed796f

1. Replace the Greek with anglicized ASCII.
2. Replace the Math. prefix with an import.
3. Fix the imports.
4. Update the tests.
   5 Add reference image to the README.

from d3-geo-projection.

my recipe for setting up the dev environment, in case it can help others: https://gist.github.com/Fil/5f8c92efadd183670ae8c63a060f03b8

from d3-geo-projection.

The difference in size is expected. I run the output images through optipng (brew install optipng on macOS) before adding them to the img folder so they are smaller to download. The test uses imagemagick to compare just the pixel values, not the non-image information.

from d3-geo-projection.

Also re. your Gist, you can simplify slightly and just npm install because all those things are listed as dependencies (or devDependencies) in the package.json.

from d3-geo-projection.

I've just pushed a PR claiming winkel3. Incomplete but up for comments.

tests were failing on my Mac due to an old version of compare that was on the path (/Applications/MAMP/Library/bin : ImageMagick 6.8.9-7); I've replaced it by ImageMagick 6.9.4-9 Q16 and now they work.

Re: npm install, I thought I did just that, but I didn't (my mistake).

One question: how do you choose and specify the default scale? Just so that it fits well in the 960/500 reference canvas?

from d3-geo-projection.

Just so that it fits well in the 960/500 reference canvas?

Yes; typically with a 5-10px margin. Also pick suitable clipping, since not every projection wants antimeridian clipping.

from d3-geo-projection.

I’ve started trying to do a mass translation over here:

https://github.com/d3/d3-geo-projection/tree/4-broken

from d3-geo-projection.

The game looks simpler now 🤖. Please give the new rules.

from d3-geo-projection.

from d3-geo-projection.

Another question: wouldn't it be neater to .rotate([-11.5,0]) by default on most projections so that we cut in the waters?

from d3-geo-projection.

Hmm. I can see the appeal in that, but I also think it’s best to stick with the exact standard aspect by default. Better to just suggest that as a better default aspect in the README, say in the top introduction (which needs pictures anyway!).

from d3-geo-projection.

OK. I'd keep it in a funky projection such as Wiechel (where you had put it in v3)
https://github.com/d3/d3-geo-projection/pull/38/files#diff-7aedb1c4d81a3fbd4bf980fc6e6bd479

from d3-geo-projection.

from d3-geo-projection.

The visual tests are probably sufficient to test projections, but what about inverse projections? Can we use a systematic method? Maybe test a given or random set of points?

from d3-geo-projection.

from d3-geo-projection.

HowIHateCamelCase

from d3-geo-projection.

I'm leaving the game now, not seeing straight and I think these are too difficult:

• d3.geoHealpix
• d3.geoHomolosine
• d3.geoInterrupted
• d3.geoTwoPointAzimuthal
• d3.geoTwoPointEquidistant

from d3-geo-projection.

Thanks for all your work! 👍

I merged as many as I could, but I’m on vacation now and want to spend some time with family. I’ll take another pass sometime later…

from d3-geo-projection.

I merged to master. Still have to implement the two-point and interrupted projections, but that should happen soon. And then 1.0!

from d3-geo-projection.

🎉

from d3-geo-projection.