On 22 February 2024 at 04:01, Duncan Murdoch wrote: | For you to deal with this, you should make arrow into a suggested | package,
For what it is worth, that is exactly what package tiledb does. Yet the Suggests: still lead to a NOTE requiring a human to override which did not happen until I gently nudged after the 'five work days' had lapsed. So full agreement that 'in theory' a Suggests: should help and is the weaker and simpler dependency. However 'in practice' it can still lead to being held up up when the weak-dependency package does not build. [ As for Dénes's point, most if not all the internals in package tiledb are actually on nanoarrow but we offer one code path returning an Arrow Table object and that requires 'arrow' the package for the instantiation. So it really all boils down to 'Lightweight is the right weight' as we say over at www.tinyverse.org. But given that the public API offers an Arrow accessor, it is a little late to pull back from it. And Arrow is a powerful and useful tool. Building it, however, can have its issues... ] Anyway, while poking around the issue when waiting, I was also told by Arrow developers that the issue (AFAICT a missing header) is fixed, and looking at CRAN's incoming reveals the package has been sitting there since Feb 8 (see https://cran.r-project.org/incoming/pending/). So would be good to hear from CRAN what if anything is happening here. Dirk -- dirk.eddelbuettel.com | @eddelbuettel | e...@debian.org ______________________________________________ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel
