- Oct 05, 1990
-
-
ram authored
not a subtype of the asserted type, not just when the arg *value* is nota subtype. This is necessary to prevent representation selection from being defeated.
-
wlott authored
-
wlott authored
-
wlott authored
set-vector-subtype.
-
wlott authored
Also fixed it to use XP when *print-pretty* is true.
-
ram authored
-
ram authored
-
ram authored
intersects with the combination type, so that if we later discovered that there was no intersection we wouldn't flame.
-
ram authored
any necessary efficency note
-
wlott authored
-
wlott authored
-
- Oct 04, 1990
-
-
wlott authored
structurep that is faster.
-
wlott authored
-
wlott authored
-
wlott authored
-
wlott authored
the pretty-printer can have a chance at doing something with it. Nuked Robs-backtrace.
-
wlott authored
-
ram authored
faslops so that we can reliably dump all IEEE floats.
-
ram authored
load all the IEEE special values.
-
ram authored
float for hashing float, so that SXHASH don't gag on NaNs and infinities. Also, fixed it to not EQ hash random objects, as that is illegal.
-
- Oct 03, 1990
-
-
wlott authored
-
ram authored
to do known-values return, but we can't due to variable values.
-
ram authored
use it so that we don't dump out the guts of the macroexpansion every time we get a warning attributed to the lambda.
-
ram authored
reduce the verbosity of efficency notes. These limit the number of templates flamed about, and also prevent flaming about templates that aren't significantly better than the one we ended up using.
-
ram authored
correctly.
-
chiles authored
-
ram authored
in the efficency note.
-
ram authored
failed transform, instead of just one per call. Also, when a transform quietly gives up, blow away any old note for that transform.
-
ram authored
*FAILED-OPTIMIZATIONS*.
-
ram authored
before switch demons are run (such as -edit.)
-
ram authored
as a closure rather than as a function that must be compiled for each operator.
-
ram authored
-
wlott authored
-
wlott authored
will work.
-
chiles authored
loop in truncation (trying a guess). We created a new primitive to multiply two digits and add three digits to support a more tense multiply loop. We also used this in fixnum X bignum. William fixed subtract by noticing a situation where the result needs to be one larger than the old assumption.
-
wlott authored
#|...|#'ed the :constants arg to purify, 'cause it doesn't work.
-
wlott authored
-
wlott authored
-
wlott authored
-
wlott authored
-