R-base 4.1: Patch ATSFont calls to CGFont calls, update gcc11 to gcc12

[sth0]

R-base4.1 does not compile on MacOS 13. This OS release removed the deprecated ATSFont calls for the newer CGFont calls. This replaces the calls with the corresponding new API. Also updated the link to gcc12-shlibs from gcc11-shlibs.

[dhomeier]

Built successfully on Sonoma 14.2.1 / Xcode 15.1 without the update to gcc12, which currently does not build on Sonoma (not sure about Ventura with Xcode 15).
It does build with gcc13, but that in turn may not be available on older systems any time soon – see #1080 (comment) .
I have previously tested a basically identical patch on Big Sur and High Sierra with gcc11, the patch seems to be OK back to 10.13 at least. So I'd recommend to stick with gcc 11.4 for the moment.

[nieder]

R packages that use gccXX hardcode the version that R-base was built against. Which is why you see lines like this:

fink-distributions/10.9-libcxx/stable/main/finkinfo/libs/rmods/cran-tidyr-r.info

Line 57 in 60fcd1b

(%type_raw[rversion] << 3.6) gcc5-shlibs | (%type_raw[rversion] = 3.6) gcc9-shlibs | (%type_raw[rversion] >= 4.0) gcc11-shlibs,

If we update r-base41 to gcc12, then we have update all relevant cran-XXX-r41 packages to use gcc12 (apart from gccXX availability issues across macOS versions).

[sth0]

Will regress to gcc11-shlibs

[sth0]

I have been using the PR #1080 #1080 which seems to allow building of gcc11 / 12 / 13 on all the systems I have tried.

[sth0]

pushed commits to regress to gcc11 compiler and bumped revision number.

[dmacks]

This patch looks like upstream's https://bugs.r-project.org/show_bug.cgi?id=18426 (in which case it's obviously a Good Solution:) so we should cite it as such in the .info (might be cleaner to have that attachment literally as a PatchFile2 instead of merged into the main PatchFile).

[sth0]

I had trouble applying it as a separate patch and thus combined it. I can try to separate them out as separate patches if the consensus is that is "cleaner".

[dmacks]

Doesn't really matter whether it is a seperate .patch or not, whatever is easier. But if that's the origin of the patch, definitely need to document that fact in the .info.

[sth0]

Will update the implementation description.

[sth0]

Added comment to BuildDesc tag and removed references to PCRE1 which is not used for R version 4.1