Depend on core 2.17.1, TileDB-Py 0.23.1, TileDB-R 0.21.1

[johnkerl]

#866

[codecov-commenter]

Codecov Report

All modified lines are covered by tests ✅

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

see 35 files with indirect coverage changes

📢 Thoughts on this report? Let us know!.

[johnkerl]

IIUC R CI
https://github.com/single-cell-data/TileDB-SOMA/actions/runs/6342945123/job/17229744703?pr=1731
is failing with

Package required and available but unsuitable version: ‘tiledb’

b/c for source we have 0.21.1
https://cran.r-project.org/web/packages/tiledb/index.html
but for binary we have only 0.21.0
https://r2u.stat.illinois.edu/ubuntu/pool/dists/jammy/main/

[r-cran-tiledb_0.21.0-1.ca2204.1_amd64.deb](https://r2u.stat.illinois.edu/ubuntu/pool/dists/jammy/main/r-cran-tiledb_0.21.0-1.ca2204.1_amd64.deb)          15-Sep-2023 15:17            20268468

[eddelbuettel]

One way is to add this again:

       - name: Install r-universe build of tiledb-r (macOS)
         if: ${{ matrix.os == 'macOS-latest' }}
         run: cd apis/r && Rscript -e "install.packages('tiledb', repos = c('https://eddelbuettel.r-universe.dev', 'https://cloud.r-project.org'))"

       - name: Install r-universe build of tiledb-r (linux)
         if: ${{ matrix.os != 'macOS-latest' }}
         run: cd apis/r && Rscript -e "options(bspm.version.check=TRUE); install.packages('tiledb', repos = c('https://eddelbuettel.r-universe.dev/bin/linux/jammy/4.3/', 'https://cloud.r-project.org'))"  

PPM did not run yesterday (unusual), and has not run today (can happen) so new official r2u binary of tiledb(-r) 0.21.1 (or any other updated CRAN packages). For tiledb-r the r-universe alternate makes a good fix.

[johnkerl]

One way is to add this again:

@eddelbuettel is there a good reason we shouldn't simply leave this in?

[eddelbuettel]

@johnkerl We could indeed. I was careful not to push it too much, not being a company resource etc pp. We would then always test with 'master branch of tiledb-r' which would not be a bad test. All this could be part of a broader discussion about how we may want to rejig the building and testing of various intertwined components here.

(What we really want is a magic wand or a more fault tolerant routine here. "Try r2u if it has the binary, else build from source (slower) or resort to r-universe" is one (not coded-up but plausible) approach. I think right we can through the fix back it because it is not part of the released package but really only part of what happens in GitHub Actions. "And what happens in GitHub Actions stays in GitHub Actions....". Otherwise lame that r2u "prod" is delayed but such is life.)

[johnkerl]

Thanks @eddelbuettel -- I think I get it now. (We've discussed such things before, but, non-R-expert that I am, it can take a while to "sink in".

We would then always test with 'master branch of tiledb-r' which would not be a bad test

I think this is not a state we want to remain in TBH. We should be testing against tagged-and-released software. This mod is only appropriate when we are testing tagged-and-released software (b/c we know within our heads) and are only working around a binary-artifact propagation delay which is (a) too long, and (b) completely out of our control.

I'll put reverting this mod on my to-do list for a subsequent PR, once the binaries have propagated.

Thanks again for explaining (again)! :)

[eddelbuettel]