On 10 December 2023 at 17:07, Simon Urbanek wrote: | As discussed here before packages should *never* set -mmacosx-version-min | or similar flags by hand.
a) That is in conflict with what was said in the past; we have used an explicit min version of 10.14 for the C++17 we were using then (and we now need a bit more so 11.0 is welcome). b) That is in conflict with how I read the R manual I quoted: R Admin, Section 'C.3.10 Building binary packages'. Recall that our package uses a binary artifact (by permission) so we have to play along and this minimum version used by both matters. c) That also appears to be in conflict with empirics. A quick search [1] at the searchable CRAN mirror finds around a dozen packages doing just that: setting a minimum version. Anyway -- I 'eventually' got the info I need. Best regards, Dirk [1] https://github.com/search?q=org%3Acran%20mmacosx-version-min&type=code -- dirk.eddelbuettel.com | @eddelbuettel | e...@debian.org ______________________________________________ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel
