MarkLinimon is doing experimental ports runs with gmake3.82. The results from the first one were pretty bad: 38 confirmed errors (5 more possible), with ~1100 ports as collateral damage, mostly from audio/portaudio and devel/p5-Module-Build.

After a brief discussion among the portmgrs, we decided to make another run with some patches to try to narrow down the collateral damage, before notifying individual maintainers. However, here's the information for those that want to get a head start on doing the work.

Note that the new 'gmake' error classification has been added to the results; OTOH, many regressions will be classified as 'makefile'. It depends on how, exactly, the build failure occurred.

notes: The table below is a union of the run results, but without the “cluster” and “truncated distfile” errors. You do not need to do anything with a package that fails with those – they are transient errors and will go away when run again. Also, “N/A” here means “it breaks elsewhere, so it is not a regression in this -exp.” “confirmed” means that with the port reverted to gmake3.81, it builds.