Since there's a cache pathname bug on Windows in 3.1.5, I recommend shipping with 3.1.5.7 (or 3.1.6, if we manage to ship faster than you do).
I would be happy to coordinate with you to release a 3.1.6 that is 3.1.5 with fixes for Allegro (and whatever other fixes come along for the ride).
As Faré reports, you will have to test with 3.1.5.7 in order to get the fix you need for Windows.
Again, I had no problem running the asdf test suite with 3.1.5.7 on Allegro / Windows using the minimakefile branch.
I suspect any problem Kevin had were related to Allegro somehow having (user-homedir-pathname) incorrectly returning NIL in their test configuration.
I committed in 3.1.5.8 another fix for uiop:run-program on ECL, see https://gitlab.com/embeddable-common-lisp/ecl/issues/149 and though it looks like to me I didn't break anything (and fixed it on MKCL, too), this requires re-testing on all platforms, particularly on Windows (sigh).
If that works for you, we could bless that as 3.1.6, and move on from there.
While re-testing, I noticed that concatenate-fasls was silently broken by LispWorks 7.0.0. Sigh. It would be nice if we fixed that for 3.1.6, but it looks like this might not technically be a regression. :-(
In any case, assuming it passes all tests on all platforms (which I haven't tested), I recommend that Franz should deliver its Lisp with the latest development version of ASDF (currently 3.1.5.8) rather than the slightly broken release 3.1.5 (notably cache location on Windows), if we don't release 3.1.6 on time.
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org The common argument that crime is caused by poverty is a kind of slander on the poor. — H. L. Mencken