The tentative plan is to replace CPS with CPS2, and to rename CPS2 to
CPS. We will add a pass to compile tree-il to CPS2, then work from
the top down to replace the CPS compiler passes.
* module/language/cps2.scm:
* module/language/cps2/compile-cps.scm:
* module/language/cps2/renumber.scm:
* module/language/cps2/utils.scm: New files.
* module/Makefile.am: Add new files to build.
* module/language/cps/intmap.scm (intmap->alist, print-intmap)
(print-transient-intmap): New functions.
Install the new functions as printers for <intmap> and <transient-intmap>.
* module/language/cps/intset.scm (make-atomic-reference)
(get-atomic-reference, set-atomic-reference!): New functions.
(*branch-size-with-edit*, *edit-index*): New constants.
(<transient-intset>): New data type.
(new-branch, clone-branch-and-set): Adapt to set edit field.
(transient-intset, persistent-intset): New exports.
(intset-add!): New interface, supporting "transient" intsets.
(intset-ref, intset-next, intset-prev, intset-fold, intset-fold2):
Work with transients.
* module/language/cps/intmap.scm (make-atomic-reference):
(get-atomic-reference, set-atomic-reference!): New helpers.
(*branch-size-with-edit*, *edit-index*): Branches now have a trailing
field, an atomic reference to their owner.
(<transient-intmap>): New record type. A mutable intmap.
(new-branch): Set the "edit" field on the branch.
(clone-branch-and-set): No editor for this field.
(assert-readable!, writable-branch): New helpers.
(transient-intmap, persistent-intmap): New exported functions.
(intmap-add!): New function.
(intmap-next, intmap-prev, intmap-ref): Work on transient intmaps.
(intmap-fold): Persist the intmap before folding over it.
* module/language/cps/intmap.scm (*branch-bits*): Switch to 32-way
branching. Marginally faster on lookup, and creation costs can be
amortized via using the transient interface. Marginal speedup for
assembler.scm compilation.
* module/language/cps/intmap.scm (meet-error): New helper.
(intmap-add, intmap-union, intmap-intersect): The "meet" argument is
optional and defaults to meet-error.
* module/language/cps.scm ($rec): Replace $letrec with $rec, which is an
expression, not a term. This means that the names bound by the letrec
appear twice: once in the $rec term, and once in the continuation.
This is not very elegant, but the situation is better than it was
before. Adapt all callers.
* doc/ref/compiler.texi (CPS in Guile): Incomplete documentation
updates. I'll update these later when the IL settles down.
* module/ice-9/eval.scm (primitive-eval): Factor out primitive=?
helper. Simplify compile-top-call. Add compile-top-branch for
primcall branches, so the compiler can see the specialized branch
operator.
* module/ice-9/eval.scm (primitive-eval): Lazily compile lambda bodies.
Special-case calls to top-level or module variables, and recognize
some of those calls as primcalls. In that case, emit closures with
the primcalls.
* module/language/cps/compile-bytecode.scm (optimize, compile-bytecode):
Marginally improve bootstrap memory usage by not retaining stale
copies of already-rewritten terms.
* libguile/libguile-2.2-gdb.scm (vm-frame-function-name): Don't default
to the address, as we will have better identifying info via the file
name.
(vm-frame-source): New helper.
(compile-time-cond): For some reason "else" matching wasn't working;
punt and use expressions.
(snarfy-frame-decorator): Rename from decorator, and adapt to new
version of Guile frame filter patch.
(vm-frame-filter): Adapt to frame filter changes, and fill in source
info.
* libguile/arrays.c (scm_i_print_array): Print the dimension whenever
the array is a true scm_tc7_array.
* test-suite/tests/arrays.test: Check that non-vector 1D arrays print as
expected.
* libguile/libguile-2.2-gdb.scm (make-scm-pretty-printer-worker):
(%scm-pretty-printer): Refactor to avoid printing all struct / array
fields by hinting these as arrays. The resulting print is not as
faithful to the original data, but that's probably OK.
These features depend on a patch to GDB that has not been merged yet.
If GDB doesn't support the annotator/filter interface, the code have no
effect.
* libguile/libguile-2.2-gdb.scm (compile-time-cond): New helper.
(snarfy-frame-annotator): Simple annotator that gives subr names for
snarfed gsubrs.
(vm-frame-filter): New filter that interleaves Scheme frames with C
frames on the stack when the user asks for a backtrace.
* libguile/libguile-2.2-gdb.scm: Update to work with Guile 2.2 stacks.
Represent the VM stack as a SRFI-41 stream, which will work better
with the frame filter interface. Add support for getting the vm
frame's function name, if possible.
* module/ice-9/psyntax.scm (resolve-identifier): There is a case where a
syntax object can resolve to itself. Prevent an infinite loop in that
case by continuing to resolve by name.
* module/ice-9/psyntax-pp.scm: Regenerate.
* test-suite/tests/syncase.test ("infinite loop bug"): Add a test.
* libguile/goops.h (SCM_VTABLE_FLAG_GOOPS_STATIC): Reserve the fourth
GOOPS flag to indicate that a class has static slot allocation.
* libguile/goops.c (scm_init_goops_builtins): Define
vtable-flag-goops-static for goops.scm.
* module/oop/goops.scm (class-has-statically-allocated-slots?): New
helper.
(build-slots-list): Instead of the ad-hoc checks for <class> or
<slot>, use the new helper.
(initialize): Accept #:static-slot-allocation? keyword.
* module/system/foreign-object.scm (make-foreign-object-type): Declare
foreign object classes as having static slot allocation.
* test-suite/tests/goops.test ("static slot allocation"): Add tests.
* libguile/goops.c (is_accessor_method, scm_compute_applicable_methods):
Fix regression from 51f66c9120 (2009).
Accessor methods are added on each subclass on which the slot is
present; therefore if a subclass doesn't have a method, then the
methods doesn't apply. Truly fixes#17355, unlike
583a23bf10.
* module/oop/goops.scm (compute-cmethod, compute-getter-method)
(compute-setter-method): Revert earlier changes.
* test-suite/tests/goops.test ("accessor slots"): Update for new
expectations, in agreement with Guile 1.8.
* module/oop/goops/compile.scm: Inline into goops.scm, leaving a
compatible interface stub behind.
* module/oop/goops/dispatch.scm: Don't import (oop goops compile), to
break circularities.
* module/oop/goops.scm: Move (oop goops util) include up to the top, and
import (ice-9 match).
(compute-cmethod): Move here from compile.scm. Add a special case for
accessor methods, so as to fix bug #17355.
(compute-getter-method, compute-setter-method): #:procedure slot is
now generic.
* test-suite/tests/goops.test ("accessor slots"): New test.
Fixes bug #17355.
* module/oop/goops.scm (memoize-effective-method!): Adapt to
compute-effective-method change.
(compute-effective-method, %compute-effective-method): Renamed from
compute-cmethod; now a generic protocol.
(compute-specialized-effective-method)
(%compute-specialized-effective-method): New sub-protocol.
(memoize-generic-function-application!): Adapt to call the hard-wired
compute-applicable-methods based on the concrete arguments types --
the semantics is that %compute-applicable-methods is the
implementation for <generic> functions. Perhaps we should do the same
for sort-applicable-methods and method-more-specific?.
(compute-getter-method, compute-setter-method): The standard
#:procedure is now a generic slot-ref. It wasn't valid to inline
field access here, because subtypes could have different field
layouts.
(compute-applicable-methods): Refactor generic definition to use
lexical scoping.
(compute-specialized-effective-method): New method for
<accessor-method>, which does field access inlining based on the
concrete types being applied.
* test-suite/tests/goops.test ("accessor slots"): New test.
* module/oop/goops.scm (fold-slot-slots): Add `slot-ref/raw' slot, which
is what the slot-ref slot was. Now the slot-ref slot checks that the
slot is bound, if needed.
(slot-definition-slot-ref/raw): Define.
(make-slot): Adapt. Also, effective slot definition slots have no
initargs.
(define-standard-accessor-method, bound-check-get, standard-get)
(standard-set): Move definitions up.
(allocate-slots): Adapt. If the slot has an init thunk, we don't need
to check that it's bound.
(slot-ref, slot-set!, slot-bound?): Simplify.
(class-slot-ref): Use the raw getter so that we can call
`slot-unbound' with just the class.
(compute-getter-method, compute-setter-method): Simplify to just use
the slot-ref / slot-set! functions from the slot.
(%initialize-object): Simplify.
* module/oop/goops.scm: Rewrite generic function dispatch to use chained
closures instead of compiling specific dispatch procedures. The big
speed win before was not allocating rest arguments, which we gain by
simply pre-generating dispatchers for arities of up to 20 arguments.
Also now a tail call without reshuffling arguments -- which is what
dispatch now is -- is just a (mov 0 new-procedure) and (tail-call),
which is pretty cheap.
(%invalidate-method-cache!): Use the new
recompute-generic-function-dispatch-procedure!.
(arity-case, multiple-arity-dispatcher, single-arity-dispatcher)
(single-arity-cache-dispatch)
(compute-generic-function-dispatch-procedure)
(recompute-generic-function-dispatch-procedure!): New internal
interfaces.
(memoize-effective-method!): Update for new interfaces.
(memoize-generic-function-application!): Rename from `memoize-method!'.
* module/oop/goops/describe.scm (describe): Remove commented code.
* module/oop/goops/save.scm (get-set-for-each, access-for-each): Update
these hoary routines for the new <slot> universe.
