3793cf676c
implementing stack traces
...
some cleanup
2009-07-06 03:56:48 +00:00
494e439510
using lisp value stack for call frames instead of the C stack
...
adding the ability to grow the value stack as needed
the net effect is that calls use much less space, and stack frames
can use all available heap space. the only downside is that C builtins
must be aware that the stack can change out from under them if they call
lisp code. currently the only example of this is table.foldl.
also fixing bug where exceptions failed to unwind the gc handle stack.
2009-07-03 18:43:15 +00:00
2f78b407ea
some renaming (intern is now symbol) and moving stuff around
...
adding scheme aliases
2009-06-30 03:21:41 +00:00
3844191d70
adding the ability for the VM to use computed goto where available
...
with gcc v4.3.2 I found the combination of -O2 and computed goto to give the
best performance; with -O3 performance was a little worse and computed goto
was a wash.
2009-06-28 19:47:11 +00:00
b5897e0ad1
adding maxstack calculation and stack overflow protection
...
making append benchmark more useful by measuring interpreter
speed instead of the builtin
2009-06-27 23:07:22 +00:00
8decdc4ae4
making append a builtin
...
making branch destinations relative so we can access the code
using *ip instead of code[ip]
doing the same to bp (argument access)
small improvement to map, builtin->instruction
better profiling
2009-06-15 02:25:21 +00:00
6ec69f5375
optimizations to map and case
...
several optimizations to the compiler itself, mostly involving
emit and encode-byte-code
experimental, very basic profiling utility
2009-06-11 00:34:50 +00:00
943c0d71d1
tiny performance tweaks: removing some variables, making some static
2009-06-09 18:46:02 +00:00
27a3e413d3
adding opcodes loada0, loada1, loadc00, loadc01
...
adding offset and count arguments to dump
2009-06-08 03:22:32 +00:00
17752decf4
implementing support for lambdas with up to 2^32-1 required arguments
...
using copy-list to implement append2
fixing bug in op_vargc when nrequired==MAX_ARGS
making fetching of multibyte numbers from bytecode portable
some cleanup in compiler
2009-06-06 21:15:54 +00:00
7e59898778
updating doc
2009-05-31 22:27:37 +00:00
ba32e4b0e9
simplifying the whole list* situation, taking better advantage of
...
existing builtin functionality
2009-05-31 18:58:09 +00:00
7e65db3e74
faster append
...
removing 'equal' alias
removing some top level bindings
2009-05-30 21:13:13 +00:00
bbcc68cfdf
making copy-list a builtin, since the functionality was there anyway.
...
adding builtin primitive apply-nlist*, to speed up list* and nlist*
2009-05-30 17:54:27 +00:00
bfbbf051c9
doing a better job hashing circular structure. the hash function is now
...
always fast and gives conservative correct answers, at the expense of
fidelity on medium and large size structures. for example (hash (iota 15))
gives the same result as (hash (iota 14)).
2009-05-20 03:39:20 +00:00
c2026ba77c
adding gc handles, making evaluator stack static
...
this provides a better interface and could only help performance
starting to add some useful library code
2009-05-19 02:54:56 +00:00
c38c47d264
adding R6RS div, mod, div0, mod0
...
small improvement to cmp_eq
explicit -fomit-frame-pointer was causing test failures with gcc4.3.2
2009-05-14 17:54:59 +00:00
e365cb1d33
adding divide by 0 check in div
2009-05-14 02:53:04 +00:00
ae2a4dd156
adding integer division function div
...
changing / to do real division always
fixing equal? to distinguish -0.0 and 0.0
making equal? and eqv? false on numbers of different exactness
2009-05-14 01:30:25 +00:00
6c549f0399
fixing hash of circular structure to avoid tree recursion (very important!!)
...
simplifying vector_grow
2009-05-13 04:03:13 +00:00
77e37368c9
fixing trace and untrace
...
replacing function->vector with function:code, function:vals, and
function:env
2009-05-13 01:13:40 +00:00
a7a31cf53a
fixing vararg apply to work properly
2009-05-11 23:55:40 +00:00
c94774a5df
adding *error-stream* variable
...
removing unnecessary byte after :let in bytecode encoding
adding script for easy bootstrapping
2009-05-05 23:51:13 +00:00
63edc82ba4
using global variables uniformly for print settings instead of passing
...
around one of them
fixing unwind-protect not to duplicate code
2009-05-05 04:01:06 +00:00
264df1f90b
improving closure representation
...
some performance tweaks
2009-04-28 04:10:18 +00:00
14d625bd83
some performance tweaks
2009-04-27 03:21:53 +00:00
de19e4f401
adding return special form
...
eliminating some unnecessary top-level bindings
adding opcodes add2, sub2, neg
2009-04-26 22:19:32 +00:00
0dc4c16276
making procedure? a builtin
...
fixing type error trying to apply #t, #f, nil
2009-04-24 20:43:56 +00:00
909b91ffcc
fixing printing of invalid UTF-8
2009-04-23 00:22:03 +00:00
e4488bb065
some tweaks to the vm
2009-04-22 23:00:13 +00:00
27120b0ce4
fixing bug in truncate
...
simplifying nconc loop
2009-04-22 15:02:49 +00:00
d6f73d0bf0
implementing #vu8 read and print syntax for uint8 vectors
...
much more compact string representation of bytecode
2009-04-21 15:35:46 +00:00
305c350c56
a couple small fixes
...
making bytecode representation more space-efficient
2009-04-21 14:53:18 +00:00
25812731cc
eliminating interpreter. the bytecode VM is now fully bootstrapped.
...
making the empty vector a singleton
removing syntax environment stuff from core
reimplementing eval using the compiler
fixing a couple bugs in long argument lists
2009-04-21 00:56:05 +00:00
5ab7a7c1e1
adding new "translucent" function type for byte-compiled lambdas
2009-04-19 22:22:17 +00:00
86b7738c89
cleaning up implementation of apply() entry point
...
removing use of interpreter in computed calls to builtins
2009-04-17 03:40:52 +00:00
94814a2e34
a bug fix and a first pass at let-optimization
2009-04-16 21:20:15 +00:00
0a3590aa01
some optimizations
2009-04-16 14:21:16 +00:00
ad4a086790
converting for to a special form
...
adding loadi8 instruction
cleaning up numeric comparison, reducing repeated code
2009-04-16 03:05:38 +00:00
36a209cd5f
making = a builtin
...
fixing = and eqv? to work properly on NaNs
fixing other comparison predicates to be consistent
2009-04-15 23:54:43 +00:00
81641a2240
fixing glaring bug in last revision
2009-04-15 00:30:24 +00:00
672558d30f
bytecode vm is now working, off by default
...
various bug fixes
language changes:
• constant symbols no longer shadow everything
• eval* removed
• vararg lists always allocated on entry, dotted argument
lists not preserved
new applyn() entry point
2009-04-15 00:12:01 +00:00
b9a1be78a0
implementing op_closure, fix to loadc/setc
2009-04-09 16:09:02 +00:00
debf3fd517
moving (length) out of core
...
changing another recursive call to goto
adding special cases in compiler for 0 and 1 argument versions of some
vararg builtins
beginning implementation of bytecode interpreter
2009-04-09 04:04:27 +00:00
ea5d334626
some cleanup, removing some unnecessary global bindings
2009-04-08 18:17:02 +00:00
43cb51f640
replacing a recursive call with a goto; saves lots of stack space.
2009-04-07 15:55:13 +00:00
e119a66bcd
moving lognot out of core
...
fixing type check in ash
2009-04-02 03:53:38 +00:00
43e8d1fbf0
adding the ability to heap-allocate extentions to the value stack,
...
so recursion depth is limited only by the process stack limit.
reorganizing evaluator so the same code is used for evaluating
and pushing arguments for both builtin functions and lambdas.
for now this is slower, but it was done in preparation for
Things To Come.
adding list-head
implementing the calling convention for long argument lists in
bytecode compiler. arguments are broken down into a nest of
list and nconc calls.
also implementing vararg builtins.
2009-04-01 04:31:49 +00:00
4cb9685266
adding support for arbitrarily-long argument lists
...
argument lists are heap-allocated after a certain cutoff (currently 127)
2009-03-27 03:06:55 +00:00
fe72c101e2
avoiding sprintf for error messages where possible
...
moving raise, logand, logior, logxor, and ash out of core
changing prog1 to a special form
2009-03-25 02:28:21 +00:00
JeffBezanson