make clean validate html-web test exemplars from the released branch on 5 different machines. This was not a rigorous test, more of a semi-controlled casual comparison. The main failure for good control was that (in
some cases) there were different error messages, because different systems fail to run certain things in different ways. But on quick check by eye, in no case was this a large contributing factor. (This could be rectified by doing it right, i.e. by doing it
in the docker container on each machine. But several of these machines are too old or under-resourced to run docker. I will probably do the docker test on the 3 fastest of them some day.) The other control failure is that although I quit most other running
programs (except, of course, Emacs, as I was running the build in Emacs, and did not check that it was the same version of Emacs), I did not check for running background daemons, etc.