It's working reliably for me for Allegro (and modern) on Linux. Sigh. Maybe sleep 1 is not reliable on these systems, and I need to sleep for 2 seconds? Nope, doesn't work for me.
I hate timing-dependent tests. Maybe I should make it time-independent for ASDF using the cache frobbing like other tests, and only resort to time-dependencies when testing native defsystems? But then, is it still a fair comparison between defsystems?
I'm too tired to work on that anymore for tonight at least. It's a new test, maybe it can be disabled for now, at least for select OS/implementation combinations, with a #+(or ...) (leave-test "Mysteriously fails on ... fixing later" ...)
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org To make an apple pie from scratch, you must first create the universe. — Carl Sagan
On Mon, Oct 14, 2013 at 10:08 PM, Robert P. Goldman rpgoldman@sift.info wrote:
Faré wrote:
The ABCL test fails reliably on Linux, too, from the command-line. But when I run it interactively at the REPL, it works reliably (!). That's puzzling, to say the least, and would require more energy than I have left at the moment. Could we mark that as a known failure for ABCL for now?
Hm. Looks like Dave Cooper is having the same error on alisp on Windows.
Is there any chance we have a problem where the make script is causing something odd to happen with the filesystem?