search

dankelley / ocedata

oceanographic data to support oce package
https://dankelley.github.io/ocedata
8 stars 2 forks source link

dataset-loading problem #11

Closed dankelley closed 5 years ago

dankelley commented 5 years ago

I just received the following, and will look into this asap.

Quote

This concerns

HydeNet NatureSounds dsm hglm.data ocedata primerTree ragt2ridges rbugs wrspathrowData

As you will see from the latest R-devel check results, your packages contain at least one dataset which cannot be loaded without an undeclared package being installed. ('At least one' as the check stops at the first missing package.)

Given that usage such as data("foo", package="bar") is allowed, it is best to eliminate such dependence if possible. Otherwise you need to declare the missing package(s) in Suggests.

There are a some known causes of such dependence:

(a) saving a copy of a dataset which is (or a component is) of an S4 class declared in another package (such as "sp" or "RCurl").

(b) saving a function (and hence its environment) generated by another package -- in several past cases the environment could be reset.

Please correct ASAP and before Jan 16 to safely retain the package on CRAN. (Be aware of the CRAN shutdown from Dec 21 to Jan 2 inclusive.)

-- Brian D. Ripley, ripley@stats.ox.ac.uk Emeritus Professor of Applied Statistics, University of Oxford

Endquote

dankelley commented 5 years ago

Oh, this is an 'ocedata' problem, actually. I'll move it.

dankelley commented 5 years ago

It is 'oce' that is missing; e.g. https://www.r-project.org/nosvn/R.check/r-devel-linux-x86_64-fedora-clang/ocedata-00check.html contains:

checking data for non-ASCII characters ... WARNING
  Error in .requirePackage(package) : unable to find required package 'oce'
  Calls: <Anonymous> ... .extendsForS3 -> extends -> getClassDef -> .requirePackage
  Execution halted
  The datasets use package(s) not declared in the DESCRIPTION file.

My plan is to try to tackle this using rhub builds, before trying CRAN submissions. That should lead to fewer hassles.

I am going to try the following (a malleable list that I will edit in-place, also writing notes below items to report on my results).

dankelley commented 5 years ago

I feel that the best choice is to add 'Suggest:oce' to DESCRIPTION, write a specific cran-submit note, and then just submit it to CRAN, adn see what happens. CRAN is only open for another day before the solstice shutdown (and I'm guessing European timezones on that), so if I do this today, we may see if it will work. (I suspect that the backgound checks will still happen in the CRAN build system, so that if this creates a circular 'suggest' dependency with oce/ocedata, we'll find that out before the few days in January before the deadline).

Unless I hear from other devels (or readers) by 10AM Halifax time, I will go ahead and do this.

dankelley commented 5 years ago

Git commit c51ae954e9dc20fc82bf5557c8709cb908209662 holds the update that I plan to try submitting to CRAN at noon, Halifax time. The submission often has snags that take an hour or two to smooth out, but this ought to let me get it submitted by late afternoon England time, meaning that we may be able to get feedback on this before CRAN goes offline for the solstice holiday. I would love to have this cleared up before the Jan 16 deadline, because if ocedata gets removed from CRAN, a lot of oce users will find themselves scrambling for finer-resolution coastline datasets.

dankelley commented 5 years ago

I just submitted it to CRAN, using devtools::release(). Note that this action required updating RStudio and several packages that do extra testing ... I won't bother making notes on the procedure, because it is all directed via a list of questions that release() asks, and likely those questions will be just a bit different in a month's time.

I will close this issue if (when?) ocedata gets through the CRAN hurdles, hopefully within a day but otherwise it will be after the CRAN solstice break.

dankelley commented 5 years ago

Step 1 passed:

Dear maintainer,

thanks, package ocedata_0.1.5.tar.gz is on its way to CRAN.

Best regards,
CRAN teams' auto-check service
Flavor: r-devel-linux-x86_64-debian-gcc, r-devel-windows-ix86+x86_64
Check: CRAN incoming feasibility, Result: Note_to_CRAN_maintainers
 Maintainer: 'Dan Kelley <Dan.Kelley@Dal.Ca>'
dankelley commented 5 years ago

According to https://cran.r-project.org/web/checks/check_results_ocedata.html this update seems to be doing fine, having passed for two systems. I will keep this issue open until I notice that it has passed on all systems, which I suppose may take a few days. In the meantime, I see that this version is now listed as the official version on https://cran.r-project.org/web/packages/ocedata/index.html

dankelley commented 5 years ago

All checks have now passed (see https://cran.r-project.org/web/checks/check_results_ocedata.html) so I'm closing this issue now.