bioconda-recipes
bioconda-recipes copied to clipboard
bioconda
Bioconductor Release 3.15
Now that R 4.2 is up on conda-forge, we should be able to make this happen. @dpryan79 any way I can help update the recipes?
@mjsteinbaugh The next step is to convince conda-forge to run the CRAN migration, since we need that (more or less) finished as the last step before we can start. I just merged in R 4.2.1 on conda-forge, so I'll make a PR on conda-forge-pinning and request a migration.
Is there any update on this? It would be helpful to know what the blockers are (presumably on the conda-forge side) so that we can help out if needed.
AMAZING! Thanks for letting me know. This will allow us to comment out so many docker containers in our workflows.
@dpryan79 Following up on this, should we wait and just build Bioconductor 3.16 instead? It is scheduled for release on November 2nd.
https://bioconductor.org/developers/release-schedule/
I'd vote to build both. Bioconductor is too core to bioinformatics users to have a gap in versions.
The Conda Forge R 4.2 migration is mostly done, BTW.
We'll skip over 3.15. It takes about 2 weeks to build all ~3000 Bioconductor packages and by then they will have been deprecated. At this point there's no real gain to burning many hundreds of hours of CI resources.
One question, if there would be a stable bulk build on all bioconductor release-XXX packages.
All bioconductor-XXX bioconda packages would be redundant and one should think about to remove this kind of R-package branch from bioconda?
I am just on the way to include my R-Bioconductor package to bioconda, mainly to make it accessible to Galaxy afterwards. If there is a stable bulk bioconductor biconda release, at least this type of integration should be covered to deal with downstream pipeline integration.
Hi, it is MSA2dist the problem for the Bioconductor release-3.16 so far is that the most used package/dependency Biostrings, so far is failing for OSX. With this issue using an OSX as the working OS, it is not possible to use the common approach (https://bioconda.github.io/contributor/guidelines.html) to include any other Bioconductor package that depends on Biostrings.
see e.g. https://github.com/bioconda/bioconda-recipes/issues/37827 https://github.com/bioconda/bioconda-recipes/issues/37535 https://github.com/bioconda/bioconda-recipes/issues/37224
OSX builds to take usually longer (we have simply more Linux capacity) than Linux builds. And the current 2.66 Linux build is just a day old, so I would be patient. I assume that the current bulk build of all packages will still take a week until most packages are build, so if you can wait a week, we can look at those issues again.
Also keep an eye on https://anaconda.org/bioconda where you see the latest packages arriving.
@bgruening is the release of Bioconductor 3.16 expected to break packages for older releases (such as 3.14 and R 4.1)? Because I'm running into errors with packages that were installing fine a couple of days ago: https://github.com/bioconda/bioconda-recipes/issues/37906
The irony is that the package version that I'm referencing there, bioconductor-org.hs.eg.db-3.14.0-r41hdfd78af_1, was released a couple of months ago to fix these exact same errors with post-link.sh, which at that time were happening with the previous iteration of the package (bioconductor-org.hs.eg.db-3.14.0-r41hdfd78af_0), which was released one year ago.
https://anaconda.org/bioconda/bioconductor-org.hs.eg.db/files
@amizeranschi The new packages won't affect old ones. There's something very strange going on here.