I suppose you're talking about compute-action-stamp in plan.lisp. Indeed, there is a bug: When in ASDF 3.3 I swapped the timestamp meaning of NIL and T, I failed to change the condition in missing-in from missing-out from :unless s to :when (eq s t)
There ought to be tests for that. (Not volunteering to do the hard work, sorry.) —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org “The first lesson of economics is scarcity: there is never enough of anything to fully satisfy all those who want it. The first lesson of politics is to disregard the first lesson of economics.” — Thomas Sowell On Sat, Feb 1, 2025 at 12:52 PM Robert Goldman <rpgold...@sift.info> wrote: > > The following message was received to the mailing list and, for some reason, > discarded instead of held for moderation (if you want to submit emails to > asdf-devel, please subscribe; you can always unsubscribe later.) > > This certainly seems plausibly to be a bug. Please post it at GitLab and I > will have a look at it as time permits. A test case would be a real plus -- > I'm much more likely to get on this if I don't also have to create the test > case myself. > > ________________________________ > > Hello, > > I may have discovered a bug in ASDF. > > In the function COMPUTE-ACTION-STAMP, first the IN-STAMPS and then the > OUT-STAMPS are collected. When one of the last write dates are not > available, T is returned by GET-FILE-STAMP, which is deemed infinitely > old by TIMESTAMP<=. > > But suppose that all the IN-STAMPS and at least one of the OUT-STAMPS > are not available, then LATEST-IN and EARLIEST-OUT are both T, and the > wisest thing would probably be rebuild everything, but TIMESTAMP<= > returns T so nothing is built. > > What do you think? > > Andrea Monaco