On Fri, Feb 22, 2013 at 8:42 PM, Dave Cooper david.cooper@genworks.com wrote:
I must be missing something because I thought Quicklisp already rejects at least new dists if they throw warnings on compiling. Maybe they were/are some different class of warnings, but I remember clearly that Genworks got rejected for inclusion in Quicklisp (this was just over a year ago, and we got "cantbuild" labels in our Issue in quicklisp-projects) until we got all Warnings eradicated.
In the meantime it looks like a few more mis-ordered defparameters crept in, which did not cause new Quicklisp rejections but were flagged by Anton's recent CL-test-grid results with tnew strict asdf.
So if the Quicklisp policy has been like this since the beginning, that seems to imply that any warnings in current systems have come in sometime since the system was first accepted into Quicklisp.
Quicklisp hasn't changed here. It relies on ASDF to build software. It's ASDF that has been catching the warnings, and on SBCL failing the build if there is any non-style-warning warning, as per asdf:*compile-file-failure-behaviour* (the default value of which varies by implementation). Now, what warnings get detected evolves as implementation are improved, or in this case as ASDF itself gets more clever and checks deferred warnings that it previously was failing to check.
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org Democracy is but government of the busy, by the bully, for the bossy. — Arthur Seldon, "The Dilemma of Democracy"