Interested in the development of ggOceanMaps?

[MikkoVihtakari]

Not sure whether this is the right place. Please close the issue, if not.

I was wondering whether some of you in the (gg)oce community would be interested in the development of ggOceanMaps. That package could be used in ggoce as the mapping engine. By being involved in the development, you would have a say in which direction to develop the package. Currently, I am developing the package alone. I notice that it would be nice to have a couple of more persons on the team.

I have been thinking about rewriting the package to use sf instead of sp. At the same time, one could consider redesigning the plotting philosophy to support a more flexible definition of crs ala. coord_sf().

[richardsc]

I'm going to tag @paleolimbot here, since he is the primary developer of ggoce and also an sf super-whiz (and I'm not sure he follows oce issues, but of course will see the accompanying ggoce issue).

@dankelley can comment on the "sp-to-sf" transition, since we did it for oce recently (I was mostly only involved in the testing, not the actual code-rewriting).

[dankelley]

The oce transition was from (sp, rgdal) to sf. I had several reasons for this.

1. To avoid using multiple packages. The more packages, the more difficult to code.
2. For a while, it seemed that rgdal was planning to drop a function oce needed.
3. It seems that development momentum is shifting from sp to sf.

I found the developers of all these packages to be very patient and helpful, as I stumbled along in the conversion process. From my experience, I am guessing that you'll find a transition to sf to be relatively painless.

I cannot comment much on the gg aspect, because I live in base-R space. @paleolimbot knows a great deal about that, and spatial work, though.

[paleolimbot]

Definitely a +1 to move to sf! It's where development is focused now.

With respect to ggoce, I don't think it needs a mapping engine at all...there are a few functions that turn oce objects into tibbles and that's all that's needed to feed them into ggplot or ggOceanMaps. With respect to development effort, I'm more than happy to continue maintaining ggspatial to make sure it continues to serve ggOceanMaps well! Beyond that, I've already committed myself to more projects than I have time to contribute to, but feel free to flag me whenever you think I can add to the discussion!

[MikkoVihtakari]

Thanks for the helpful comments :) If the upcoming ggplot2 update breaks everything as it looks now and ggOceanMaps (and ggspatial) get taken out of CRAN for a while, I think I'll do the change this summer. Otherwise, I'll wait with it until I have a better time.

Perhaps we can leave this issue open for a while. Maybe there are some other devs interested who also have time (lack of time and too many commitments is the status quo for most of us).

I have been dabbling with the shift many times but always gave up because I found out that direct translation from sp to sf would make ggOceanMaps very slow. That is probably because I do not use the sf code optimally. Here is one example: r-spatial/sf#1666 (comment)

[stale]

This issue has been automatically marked as stale because no comments have been made in two weeks. Unless comments are made in the next two weeks (or the 'pinned' label is applied), this issue will be closed automatically in a week.