rpy2 is built by default when a suitable system R is detected.
On Saturday, March 25, 2023 at 9:29:32 AM UTC-7 William Stein wrote: > Hi, > > According to > > https://doc.sagemath.org/html/en/reference/spkg/rpy2.html#spkg-rpy2 > > the rpy2 package is a "standard package". However, I just checked > multiple builds in various places of sage-9.8 and also the latest > master branch, and rpy2 is not installed. > > I can see why rpy2 isn't installed. It's because for sage-9.8 R was > deprecated from being a standard package to an optional package, and > of course there is no way to install rpy2 if a sufficiently library > usable R isn't installed. > > 1. What is the definition these days of a standard package? Is a > standard package "A package that might or might not be installed, > depending on what happens to be available when ./configure was run?" > A standard package used to be a package that was definitely installed > in every copy of Sage. It seems to me that these days there are three > types of packages: > > - definitely installed in every copy of sage. An example is pari. > - there's an attempt to install them when sage gets built, if > conditions are right: an example is rpy2 > - they are definitely not installed when sage is initially built: > example include various specialized databases. > > 2. A possible suggestion for improving the error messages. Just > doing "sage -i rpy2" doesn't work, because that complains that R isn't > installable: > > [...] > [r-none] If the system package is installed, ./configure will check > whether it can be used. > [r-none] > [r-none] > [r-none] Error: r is a dummy package and > [r-none] cannot be installed using the Sage distribution. > > OK, I realize I need to follow the directions above (about ubuntu), > and then I get r dev installed. I guess I know enough to > actually run ./configure again, and then I do "sage -i rpy2", and then > that does a LOT, e.g., spending hour(s) building gmp, mpfr, etc. > (why? because I ran ./configure?). > > Why does it say "If the system package is installed, ./configure will > check whether it can be used."? This isn't > user friendly. It should say: "Ensure that the system R package is > installed, then run ./configure. After you do that, you can then > run 'sage -i rpy2'." > > MOTIVATION: For every version of Sage up to and including 9.7, users > could use rpy2, which lots of little bits of sage evidently > now rely on,e g., "r.mean([3,7,9])". By deprecating the R > package entirely, thus making rpy2 into a sudo-standard package > in version 9.8, using the r interface is just broken, since it uses rpy2. > > My guess is that rpy2 and all code that depends on R should be made > optional packages and marked #optional in the docs, etc., > and it's just a bug that hasn't been done? Alternatively, we have to > make it so that building Sage depends on R being installed, or > we have to change the definition of "standard" package. > > -- > William (http://wstein.org) > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/2750bff6-475c-4818-985a-130a06a3453an%40googlegroups.com.
