CI for R ports broken by latests commits to mpbb

[Sergey Fedorov]

Could someone who understands what mpbb scripts are doing address this?
https://trac.macports.org/ticket/68408

We got numerous R ports broken by this on CI. (Locally everything works
fine, buildbot apparently too.)

I can’t keep updating R ports unless this is fixed. It is too much of a
meaningless time-waste to run endless CI, find out what is broken, exclude
those updates from PRs, run again.
Everything was working pretty much perfectly for many months until recently.

P. S. Initially I thought the issue was caused by a sudden migration to
libgcc13 – but apparently that was a coincidence. Every failure either
complains about non-existent cyclic dependencies (possibly confusing test
deps with build deps, at least in some cases) or about activation of some
dependency with a non-transparent message, pointing to mpbb.
Even some ports unrelated to R failed in this manner: for example, libiconv
(which I did not touch at all).
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.macports.org/pipermail/macports-dev/attachments/20231010/45dcc489/attachment.htm>