+Thu May 27 08:19:00 1982 by jkf
+ fixed bug in d-fixnump which caused liszt to assume too often that
+ arguments are all fixnums
+ (now: version 8.08)
+
+Wed May 12 13:46:03 1982 by jkf
+ new declaration scheme died when the object being declared
+ was a list. e.g (declare (fixnum (foo fixnum flonum)).
+ Fixed declare-handler so that cases like this are ignored.
+ (now: version 8.07)
+
+Fri May 7 19:28:04 1982 by jkf
+ allow (function (nlambda ()...)) and (function (lambda () ..))
+ (now: version 8.06)
+
+Thu May 6 15:03:50 1982 by jkf
+ support local declarations. They are stacked on g-decls.
+ localf and macarray don't fit in too well, they should
+ be fixed.
+ fixnum declarations now have meaning. They will allow the
+ compiler to convert from functions like 'add1' to 1+.
+ 'declare' is now a function in the compiler so declarations
+ may be fasl'ed in.
+
+Thu Apr 22 20:48:25 1982 by jkf
+ #' returns a symbol now, instead of the getd of the symbol.
+ This was required for maclisp compatibility.
+ (now: version 8.05)
+
+Wed Apr 21 07:50:41 1982 by jkf
+ load in init file before compiling. look in . first, then $HOME.
+ look for .liszt.o, .lisztrc.l, lisztrc.o, lisztrc.l
+
+ Upon interrupt, remove temp file and exit
+ (now: version 8.04)
+
+Mon Apr 12 23:22:37 1982 by jkf
+ open code 'function'. Now it will cause another function to be
+ generated and compiled and the function call returns the
+ bcd header for the newly created function
+ liszt-process-forms may be filled with forms to compile while
+ compiling a form. Unlike liszt-eof-forms, liszt-process-forms
+ are compiled at the next opportunity.
+ (now: version 8.03)
+
+Wed Mar 31 08:47:46 1982 by jkf
+ fixed bug in which (> (or nil 9) 0) would return true. cc-eq
+ now rebinds g-trueop and g-falseop. (now: version 8.02)
+
+Wed Mar 31 08:24:27 1982 by jkf
+ added a new file to liszt: lversion.l It is not sccsed and it
+ just contains the version number. This will allow the version
+ number to be changed with ease whenever a modification is made.
+ start it at 8.01 (now: version 8.01)
+
+Fri Mar 19 11:17:12 1982 by jkf
+ did distribution. (now: version 8.00)
+
+Fri Feb 19 09:56:50 1982 by jkf
+ fixed e-docomment so that newlines in the comment will not
+ cause assembler problems [io.l]
+
+Wed Feb 17 12:46:24 1982 by jkf
+ to version 8.0 (to correspond with Opus 38)
+
+Wed Feb 10 21:28:41 1982 by jkf
+ fixed a bug in the return function and a possible bug in go.
+ both bugs have to do with returning or going through a
+ catch or errset, which is a pretty strange thing to do anyway.
+ The fix was to keep track of the number of catches and errsets
+ before the prog to be returned from or gone into.
+ I modified d-pushframe to always push 3 args, so we can know
+ how long each frame we pop will be.
+Sat Dec 5 11:58:36 1981 by jkf
+ open coded boole (mods to decl.l and funa.l). used the open
+ coding capability of fixnums.
+ Added three internal functions fixnum-BitXor, fixnum-BitAndNot
+ fixnum-BitOr, which are the three functions which the vax
+ can do.
+Mon Oct 26 21:15:36 1981 by jkf
+ fixed bug in d-supercxr (funa.l) which made this function
+ (cond ((setq y (cxr n zip)) (print 'hi))
+ (t (print 'lo)))
+ always prints 'hi'. The problem was that only jump on true was
+ checked if the value of a cxr was stored somewhere. In the case
+ above we want to jump on nil.
+
+Sat Oct 24 16:41:03 1981 by jkf
+ -*- to version 7.1 -*-
+ Added the 'a' option to liszt, which if set will cause a
+ special form to be output after a function is defined. that special
+ form will put a 'fcn-info' property on the function's property
+ list. the form of the fcn-info property is
+ (argdesc compileinfo)
+ argdesc may be nil or (x . y), the later meaning that there are a
+ minumum of x args required and a maximum of y. Currently lexprs
+ just put 'nil' there, but soon I will introduce a way to declare
+ min and max args for lexprs via a declare. The compileinfo
+ is a string saying what file this came from and when it was
+ compiled.
+
+Wed Oct 21 20:19:53 1981 by jkf
+ added functions <&, =&, and >& which are fixnum only
+ versions of <,= and >.
+
+Tue Oct 20 22:14:41 1981 by jkf
+ fix < and > compiling so that they are only open coded if
+ we can be sure that both operands are fixnums. For now this
+ amounts to checking that one of the operands is a fixnum.
+ We should augment to this check also for type declarations.
+
+Sat Oct 17 11:47:50 1981 by jkf
+ since most fixnums are in the range 0 to 1023 we can probably
+ win by doing inline checking in that range. Thus I've added
+ the function d-fixnumbox which does an inline reboxing if
+ the number is between 0 and 1023 and otherwise calls qnewint.
+ The file changed was fixnum.l
+
+Thu Oct 8 23:35:34 1981 by jkf
+ added cc-= to open code = if possible. It looks for a fixnum
+ as one of the arguments and converts the = expression to
+ (eq (cdr ..) (cdr ..)) which should do the trick.
+
projects / unix-history / commitdiff