Fix a regression in APPEND type derivation.
authorStas Boukarev <stassats@gmail.com>
Fri, 31 May 2013 12:27:11 +0000 (16:27 +0400)
committerStas Boukarev <stassats@gmail.com>
Fri, 31 May 2013 12:27:11 +0000 (16:27 +0400)
commit8a19ff566e3a1a43cb3b2d11d2781a1c89981f43
tree3245304e1bc2d7443f36bb867d0a6a24aec1d58c
parent2378406d6eda78090dfe05e372438495aeace5e0
Fix a regression in APPEND type derivation.

Do not just test the arguments for validity by being subtypes or
supertypes of LIST, but also if it's a subytpe of NULL, since ATOM or
SYMBOL won't satisfy the LIST tests.

Regression since f32ee7d.."Better type derivation for APPEND, NCONC,
LIST.", reported by Eric Marsden.
src/compiler/srctran.lisp
tests/compiler.pure.lisp