TODO for CRS changes

[rsbivand]

to roll out new sp and rgdal, the next things have to be done

• compare reverse dependencies against sp & rgdal released RSB GDAL 2.4.4 PROJ 5.2.0
• compare reverse dependencies against sp & rgdal devel RSB GDAL 3.0.3/PROJ 6.3.0 (st_630_303_200117.zip)
• ?check against PROJ 7 release candidates
• submit to CRAN
• contribute to sf r-spatial blog TODO for CRS changes r-spatial/sf#1244
• inform twitter

[edzer]

OK, did you make a PR against sp, or shall I do that? Do they have to be rolled out in a particular order?

[rsbivand]

No, I think you picked up my commits from my fork master and applied them to the master by 74a8780, didn't you?

[edzer]

Ah, thanks, yes, I forgot.

[rsbivand]

And 3e0497b before that (for the record).

[rsbivand]

Can we try to release sp with my fork commits of 21 January, which should have benign effects on rgdal and raster and their revdeps? Then it will be easier to move to releasing GDAL3-ready rgdal. On PROJ6/GDAL3, I'm seeing about 50 new breakages across the whole revdep tree of ~ 900 packages, mostly messing with sp::CRS/raster::crs objects internally and not using accessor functions. All warned, many have reported back that problem fixed but not released to CRAN.

[edzer]

Yes, that means releasing from my master current branch, right?

[rsbivand]

Yes, please check my PRs. Maybe an RC of sp, so that you or I can check rgdal rev. 938.

[edzer]

Thanks; merged, and I bumped the version to 1.4-0.

[edzer]

While checking, I see

* checking dependencies in R code ... NOTE
Missing or unexported objects:
  ‘rgdal::checkCRSArgs_ng’ ‘rgdal::new_proj_and_gdal’

is some version of rgdal required that is not yet on CRAN? (How) can we suppress this NOTE?

[rsbivand]

This is the circular dependency problem, sp needing unreleased rgdal needs to be on CRAN first. I think we saw this earlier (maybe an sf issue)? I think we agreed to include this as a note to CRAN-team that the NOTE would be resolved when rgdal was released?

[rsbivand]

I'm still seeing 1.3-4 after git fetch upstream - have you pushed?

[edzer]

Sorry, now I have.

[edzer]

While checking on the gdal 3.0.4/proj 7.0.0RC1/geos 3.8.0 image I see lots of these warning messages:

< 8: In showSRID(uprojargs, format = "PROJ", multiline = "NO") :
<   Discarded datum Unknown based on Bessel 1841 ellipsoid in CRS definition

with

> sessionInfo()
R version 3.6.2 (2019-12-12)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 18.04.4 LTS

Matrix products: default
BLAS:   /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.7.1
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.7.1

locale:
[1] C

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] rgdal_1.5-5 sp_1.3-2   

loaded via a namespace (and not attached):
[1] compiler_3.6.2  grid_3.6.2      lattice_0.20-38

[rsbivand]

This is as expected, they come from inside rgdal, where there are also tools to moderate the numbers of warnings. In rgdal, I have set_thin_PROJ6_warnings(TRUE) at the beginning of examples{} sections in help files where needed, but not in tests/*.R. Without the warnings, I can't check back for possible impacts in .Rcheck/-EX.Rout files, but maybe that doesn't matter.

[rsbivand]

Is sp_1.3-2 attached reliable??

[rsbivand]

I see "Discarded" in tests/fail1.Rout and tests/pass1.Rout, as well as man pages for: CRS-class, degAxis, gridlines, loadMeuse, meuse, meuse.grid, recenter-methods, spChFIDs-methods, spDistsN1, spplot. None I can see in vignettes. To turn on thinning, would need to test for rgdal, its version, and whether GDAL & PROJ are NG.

[edzer]

Sorry, this was with sp 1.4-0 (posted the wrong sessionInfo()).

[rsbivand]

1.4-0 1 NOTE with devel (excess time when run with --run-donttest --run-dontrun), so OK otherwise, and rgdal 1.5-5.

[edzer]

OK, then I will submit with the message:

• expect a NOTE from rgdal, that will go away with new rgdal which needs this sp version
• expect several packages that will break, but all have been informed timely.

[edzer]

Submitted.