- Jul 23, 1991
-
-
ram authored
-
- Jul 22, 1991
- Jul 19, 1991
-
-
ram authored
transform for %unary-round on floats.
-
- Jul 18, 1991
- Jul 17, 1991
-
-
ram authored
there isn't a free register. This should eliminate many problems with the register not finding a legal packing when there is one. Now, instead of spilling a TN for the duration of the VOP, we pick some live TN to unpack and then repack it on the stack.
-
- Jul 14, 1991
- Jul 11, 1991
- Jul 10, 1991
- Jul 09, 1991
-
-
ram authored
-
- Jul 08, 1991
-
-
ram authored
-
- Jun 25, 1991
-
-
wlott authored
is non-NIL.
-
- Jun 19, 1991
-
-
ram authored
coercion would be inserted after the ALLOCATE-FRAME VOP, which caused register save computation to get confused.
-
- Jun 15, 1991
-
-
chiles authored
Made the default for COMPILE-FILE's :error-file argument be nil. It is a dated notion and never desired to expect compilation to defaultly produce an error output file. You always compile in the editor catching the output or in a system building utility that saves all the compiler's output. Removed the conditional compilation stuff to define NCOMPILE-FILE since we no longer need that for bringing up new ports.
-
- Jun 13, 1991
-
-
wlott authored
instead of the raw (untyped) value.
-
- Jun 07, 1991
- Jun 05, 1991
-
-
wlott authored
if the created array is known to be simple. Otherwise, someone might adjust it, which would cause the type to be wrong.
-
- Jun 04, 1991
- May 31, 1991
-
-
ram authored
-
ram authored
delayed. This is somewhat questionable, but is probably OK. The real problem is that (funcall #'foo ...something that fsets foo..) doesn't do the right thing. The reason for this change is that it produces much better code for notinline references (in particular, those resulting from calls the DEFSTRUCT predicate functions inside of the structure-typep transform.)
-
ram authored
-
- May 28, 1991
-
-
ram authored
well as the existing DEBUG-INFO quality.
-
- May 24, 1991
- May 23, 1991