1
Fork 0
mirror of https://git.savannah.gnu.org/git/guile.git synced 2025-05-18 18:40:22 +02:00
Commit graph

48 commits

Author SHA1 Message Date
Andy Wingo
b7946e9ec6 push new frame on stack before procedure & args
* libguile/_scm.h (SCM_OBJCODE_MINOR_VERSION): Bump

* libguile/vm-engine.c (vm_engine): Push a frame corresponding to the
  mv-call.

* libguile/vm-i-system.c: Renumber ops.
  (new-frame): New op, pushes a frame.
  (call, mv-call): No need to shuffle args, though we do need to pop the
  frame in the non-vm call case.
  (goto/args): Inconsequential tweaks.
  (call/cc): Push a frame if needed.

* module/language/tree-il/compile-glil.scm (flatten): Emit `new-frame'
  as appropriate.

* test-suite/tests/tree-il.test: Fix to expect new-frame.
2009-08-20 18:52:50 +02:00
Andy Wingo
aaae0d5ab3 "fix" <let>-bound lambda expressions too
* module/language/tree-il/compile-glil.scm (compile-glil): Compute
  warnings before optimizing, as unreferenced variables will be
  optimized out.

* libguile/_scm.h: Fix C99 comment.

* module/language/tree-il/fix-letrec.scm (partition-vars): Also analyze
  let-bound vars.
  (fix-letrec!): Fix a bug whereby a set! to an unreffed var would be
  called for value, not effect. Also "fix" <let>-bound lambda
  expressions -- really speeds up pmatch.

* test-suite/tests/tree-il.test ("lexical sets", "the or hack"): Update
  to take into account the new optimizations.
2009-08-12 21:29:08 +02:00
Andy Wingo
94ff26b96b rework the vm support for wide strings
* libguile/_scm.h (SCM_OBJCODE_MINOR_VERSION): Bump.

* libguile/vm-engine.c (vm_error_bad_wide_string_length): New error
  case.

* libguile/vm-i-loader.c (load-unsigned-integer, load-integer)
  (load-keyword): Remove these instructions. The former two are
  obsoleted by make-int64/make-uint64, the latter via make-keyword.
  (load-string): Only handle narrow strings.
  (load-symbol): Only handle narrow symbols. The wide case is handled
  via make-symbol.
  (load-wide-string): New instruction, for wide strings.

* libguile/vm-i-system.c (define): Move here from loaders.c, as now it
  just takes a sym on the stack.
  (make-keyword, make-symbol): New instructions.

* module/language/assembly.scm: Remove removed instructions. No more
  width byte in load-string etc.

* module/language/assembly/compile-bytecode.scm (write-bytecode): Adapt
  to change in instruction set.

* module/language/glil/compile-assembly.scm (glil->assembly): Compile
  define by pushing the sym then emitting (define).
  (dump-object): Dump narrow and wide strings differently. Use
  make-keyword and make-symbol as appropriate.

* module/language/tree-il/compile-glil.scm (flatten): When compiling a
  ref to a primitive (not a call), first see if the primitive is
  actually bound in the root module. (That's not the case with e.g.
  bytevector-u8-ref).

* module/system/xref.scm (program-callee-rev-vars): Don't parse out
  "nexts".

* test-suite/tests/asm-to-bytecode.test ("compiler"): Adapt to bytecode
  format change.
2009-08-12 16:34:05 +02:00
Andy Wingo
d97b69d9cd lambda, the ultimate goto
* module/language/tree-il/analyze.scm (analyze-lexicals): Rework to
  actually determine when a fixed-point procedure may be allocated as a
  label.
* module/language/tree-il/compile-glil.scm (emit-bindings): Always emit
  a <glil-bind>. Otherwise it's too hard to pair with unbindings.
  (flatten-lambda): Consequently, here we only `bind' if there are any
  vars to bind. This doesn't make any difference, given that lambdas
  don't have trailing unbind instructions, but it does keep the GLIL
  output the same for thunks -- no extraneous (bind) instructions. Keeps
  tree-il.test happy.
  (flatten): Some bugfixes. Yaaay, it works!!!
2009-08-07 19:06:15 +02:00
Andy Wingo
230cfcfb3e implement compilation of label-allocated lambda expressions
* module/language/tree-il/compile-glil.scm (flatten-lambda, flatten):
  Implement compilation of label-allocated lambda expressions. Quite
  tricky, we'll see if this works when the new analyzer lands.
2009-08-07 17:44:02 +02:00
Andy Wingo
9059993fe0 add label alist to lambda allocations in tree-il->glil compiler
* module/language/tree-il/analyze.scm: Add some more comments about
  something that will land in a future commit: compiling fixpoint
  lambdas as labels.
  (analyze-lexicals): Reorder a bit, and add a label alist to procedure
  allocations. Empty for now.

* module/language/tree-il/compile-glil.scm (flatten): Adapt to the free
  variables being in the cddr of the allocation, not the cdr.
2009-08-07 15:35:53 +02:00
Andy Wingo
9b29d60791 loop detection in the house
* libguile/vm-i-scheme.c (vector-ref, vector-set): Sync registers if we
  call out to C.

* module/language/tree-il/compile-glil.scm (flatten-lambda): Add an
  extra argument, the self-label, which should be the gensym under which
  the procedure is bound in a <fix> expression.
  (flatten): If we see a call to a lexical ref to the self-label in a
  tail position, rename and goto instead of goto/args, which will tear
  down the frame -- or will, in the future. It's a primitive form of
  loop detection.

* module/language/tree-il/primitives.scm (zero?): Expand to (= x 0).
2009-08-06 17:46:38 +02:00
Andy Wingo
80af116875 actually implement "fixing letrec"
* module/Makefile.am (SOURCES): Reorganize so GHIL is compiled last,
  along with ecmascript.

* module/language/scheme/spec.scm: Remove references to GHIL, as it's
  bitrotten and obsolete..

* module/language/tree-il.scm (make-tree-il-folder): Rework so that we
  only have down and up procs, and call down and up on each element.
* module/language/tree-il/analyze.scm (analyze-lexicals): Fix a thinko
  handling let-values.

* module/language/tree-il/fix-letrec.scm: Actually implement fixing
  letrec. The resulting code will perform better, but violations of the
  letrec restriction are not detected. This behavior is allowed by the
  spec, but it is undesirable. Perhaps that will be fixed later.

* module/language/tree-il/inline.scm (inline!): Fix a case in which
  ((lambda args foo)) would be erroneously inlined to foo. Remove empty
  let, letrec, and fix statements.

* module/language/tree-il/primitives.scm (effect-free-primitive?): New
  public predicate.
2009-08-06 16:01:24 +02:00
Andy Wingo
bca488f186 actually inline call-with-values to tree-il's <let-values>
* module/srfi/srfi-11.scm (let-values): In the one-clause case, avoid
  going through temporary variables.

* module/language/tree-il/inline.scm (inline!): Add another case:
  (call-with-values (lambda () ...) (lambda ... ...) -> let-values.

* module/language/tree-il/compile-glil.scm (flatten): Fix a bug
  compiling applications in "vals" context.

* module/language/tree-il/analyze.scm (analyze-lexicals): Fix a couple
  bugs with let-values and rest arguments.
2009-08-06 11:48:16 +02:00
Andy Wingo
4dcd84998f let-values in terms of syntax-case, add make-tree-il-folder
* module/language/tree-il.scm (tree-il-fold): Fix for let-values case.
  (make-tree-il-folder): New public macro, makes a multi-valued folder
  specific to the number of seeds that the user wants.
* module/language/tree-il/optimize.scm (optimize!): Reverse the order of
  inline! and fix-letrec!, as the latter might expose opportunities for
  the former.
* module/srfi/srfi-11.scm (let-values): Reimplement in terms of
  syntax-case, so that its expressions may reference hygienically bound
  variables. See the NEWS for the rationale.
  (let*-values): An empty let*-values still introduces a local `let'
  binding contour.
* module/system/base/syntax.scm (record-case): Yukkkk. Reimplement in
  terms of syntax-case. Ug-ly, but see the NEWS again: "Lexical bindings
  introduced by hygienic macros may not be referenced by nonhygienic
  macros."
2009-08-05 21:35:30 +02:00
Andy Wingo
c21c89b138 add <fix> tree-il construct, and compile it
* libguile/vm-i-system.c (fix-closure): New instruction, for wiring
  together fixpoint procedures.

* module/Makefile.am (TREE_IL_LANG_SOURCES): Add fix-letrec.scm.

* module/language/glil/compile-assembly.scm (glil->assembly): Reindent
  the <glil-lexical> case, and handle 'fix for locally-bound vars.

* module/language/tree-il.scm (<fix>): Add the <fix> tree-il type and
  accessors, for fixed-point bindings. This IL construct is taken from
  the Waddell paper.
  (parse-tree-il, unparse-tree-il, tree-il->scheme, tree-il-fold)
  (pre-order!, post-order!): Update for <fix>.

* module/language/tree-il/analyze.scm (analyze-lexicals): Update for
  <fix>. The difference here is that the bindings may not be assigned,
  and are not marked as such. They are not boxed.
  (report-unused-variables): Update for <fix>.

* module/language/tree-il/compile-glil.scm (flatten): Compile <fix> to
  GLIL.

* module/language/tree-il/fix-letrec.scm: A stub implementation of
  fixing letrec -- will flesh out in a separate commit.

* module/language/tree-il/inline.scm: Fix license, it was mistakenly
  added with LGPL v2.1+.

* module/language/tree-il/optimize.scm (optimize!): Run the fix-letrec!
  pass.
2009-08-05 17:52:27 +02:00
Andy Wingo
dab0f9d55d add a brain-dead inliner
* module/Makefile.am (TREE_IL_LANG_SOURCES):
* module/language/tree-il/inline.scm: Add a brain-dead inliner, to
  inline ((lambda () x)) => x.

* module/language/tree-il/optimize.scm (optimize!): Invoke the inliner.
2009-08-05 16:17:20 +02:00
Andy Wingo
7382f23e58 add1 and sub1 instructions
* libguile/vm-i-scheme.c: Add add1 and sub1 instructions.
* module/language/tree-il/compile-glil.scm: Compile 1+ and 1- to add1
  and sub1.

* module/language/tree-il/primitives.scm (define-primitive-expander):
  Add support for `if' statements in the consequent.
  (+, -): Compile (- x 1), (+ x 1), and (+ 1 x) to 1- or 1+ as
  appropriate.
  (1-): Remove this one. Seems we forgot 1+ before, but we weren't
  compiling it nicely anyway.

* test-suite/tests/tree-il.test ("void"): Fix expected compilation of (+
  (void) 1) to allow for add1.
2009-08-05 11:55:42 +02:00
Ludovic Courtès
4b856371b3 Add unused variable analysis in the tree-il->glil compiler.
* module/language/tree-il/analyze.scm (<binding-info>): New record type.
  (report-unused-variables): New procedure.

* module/language/tree-il/compile-glil.scm (%warning-passes): New
  variable.
  (compile-glil): Honor `#:warnings' from OPTS.

* test-suite/tests/tree-il.test (call-with-warnings): New procedure.
  (%opts-w-unused): New variable.
  ("warnings"): New test prefix.
2009-07-31 00:49:22 +02:00
Andy Wingo
476e357281 remove all mentions of "external" from the compiler and related code
With this, GHIL is effectively bitrotten. I need to port the ECMAScript
compiler to tree-il, then I'll remove it.

* module/language/assembly.scm (byte-length):
* module/language/assembly/compile-bytecode.scm (write-bytecode):
* module/language/assembly/decompile-bytecode.scm (decode-load-program):
* module/language/assembly/disassemble.scm (disassemble-load-program):
  (disassemble-free-vars, code-annotation):
* module/language/glil.scm (<glil-program>, <glil-local>)
  (<glil-exteral>, parse-glil, unparse-glil):
* module/language/glil/compile-assembly.scm (make-meta):
  (compile-assembly, glil->assembly):
* module/language/glil/decompile-assembly.scm (decompile-toplevel):
  (decompile-load-program):
* module/language/objcode/spec.scm (decompile-value):
* module/language/tree-il/compile-glil.scm (flatten-lambda):
* module/system/vm/frame.scm (frame-binding-ref):
  (frame-binding-set!):
* module/system/vm/program.scm (binding:boxed?):
* module/system/vm/trace.scm (trace-next):
* test-suite/tests/asm-to-bytecode.test ("compiler"):
* test-suite/tests/tree-il.test: Remove all mentions of "external", and
  of <glil-local>. Docs updates will come soon.
2009-07-23 17:15:17 +02:00
Andy Wingo
57ab0671d7 rename "closure-ref" to "free-ref"; s/vars/variables/ in some names
* libguile/programs.h:
* libguile/programs.c: (SCM_PROGRAM_FREE_VARIABLES): Rename from
  SCM_PROGRAM_FREE_VARS. Callers changed.
* libguile/programs.c (scm_make_program): Rename arg to
  "free_variables".
  (scm_program_free_variables): Rename from program-free-vars.

* libguile/vm-engine.h:
* libguile/vm-engine.c (VM_CHECK_FREE_VARIABLES): Rename from
  VM_CHECK_CLOSURE.
  (vm_engine, CACHE_PROGRAM): Rename closure and closure_count to free_vars and
  free_vars_vount.

* libguile/vm-i-system.c (FREE_VARIABLE_REF): Rename from CLOSURE_REF.
  (free-ref, free-boxed-ref, free-boxed-set): Rename from closure-ref,
  closure-boxed-ref, closure-boxed-set.
  (make-closure): Renamed from make-closure2.

* module/language/glil/compile-assembly.scm (glil->assembly): Hack to
  never write out the the old "make-closure" instruction. Will fix
  better later. Change to emit free-ref etc instead of closure-ref.

* module/language/tree-il/compile-glil.scm (flatten): Emit make-closure
  instead of make-closure2, now that the old make-closure is gone.

* module/system/vm/program.scm (system): Rename program-free-vars to
  program-free-variables.

* test-suite/tests/tree-il.test ("lambda"): Update for make-closure.
2009-07-23 17:15:17 +02:00
Andy Wingo
66d3e9a32c compile lexical variable access and closure creation to the new ops
* module/language/glil.scm (<glil>): New GLIL type, <glil-lexical>,
  which will subsume other lexical types.
* module/language/glil/compile-assembly.scm: Compile <glil-lexical>.
  (make-open-binding): Change the interpretation of the second argument
  -- instead of indicating an "external" var, it now indicates a boxed
  var.
  (open-binding): Adapt to new glil-bind format.
* module/language/tree-il/analyze.scm: Add a lot more docs.
  (analyze-lexicals): Change the allocation algorithm and output format
  to allow the tree-il->glil compiler to capture free variables
  appropriately and to reference bound variables in boxes if necessary.
  Amply documented.

* module/language/tree-il/compile-glil.scm (compile-glil): Compile
  lexical variable access to <glil-lexical>. Emit variable capture and
  closure creation code here, instead of leaving that task to the
  GLIL->assembly compiler.

* test-suite/tests/tree-il.test: Update expected code emission.
2009-07-23 17:00:56 +02:00
Andy Wingo
39141c876b bytevector ops now compile down to low-level VM ops
* libguile/instructions.c (scm_instruction_list): Fix a longstanding bug
  in this humble function.

* libguile/vm-i-scheme.c (BV_FIXABLE_INT_SET, BV_INT_SET, BV_FLOAT_SET):
  Fix some bugs in these macros -- now the bytevector ops work.

* module/language/tree-il/compile-glil.scm (*primcall-ops*): Compile
  bytevector calls to VM ops.

* module/language/tree-il/primitives.scm
  (*interesting-primitive-names*): Resolve bytevector calls to primitive
  calls.
2009-06-26 12:42:10 +02:00
Andy Wingo
d6f1ce3d16 vector-ref and vector-set! now have opcodes
* module/language/tree-il/primitives.scm
  (*interesting-primitive-names*): Resolve vector-ref and vector-set!.

* module/language/tree-il/compile-glil.scm (*primcall-ops*): And compile
  vector-ref and vector-set! to their opcodes.

* libguile/vm-i-scheme.c (vector-ref, vector-set): New opcodes, placed
  before the bytevector ops. The renumbering shouldn't affect anyone,
  given that the bytevector ops were not yet used. Fix a few bugs in the
  bytevector ops.
2009-06-26 12:42:10 +02:00
Andy Wingo
60ed31d28b allow primcall ops to push 0 values
* libguile/objcodes.c (OBJCODE_COOKIE): Bump the objcode cookie. We'll
  be doing this on incompatible changes until 2.0.

* libguile/vm-i-scheme.c (set_car, set_cdr, slot_set): These
  instructions don't have natural return values -- so declare them that
  way, that they push 0 values.

* module/language/tree-il/compile-glil.scm (flatten): When compiling
  primitive calls, check `(instruction-pushes op)' to see how many
  values that instruction will push, and do something appropriate,
  instead of just assuming that all primcall ops push 1 value.
2009-06-24 15:14:00 +02:00
Neil Jerram
53befeb700 Change Guile license to LGPLv3+
(Not quite finished, the following will be done tomorrow.
   module/srfi/*.scm
   module/rnrs/*.scm
   module/scripts/*.scm
   testsuite/*.scm
   guile-readline/*
)
2009-06-17 00:22:09 +01:00
Andy Wingo
ac4d09b164 a start to changing VM scheme copyrights
* module/language/tree-il/primitives.scm: Change copyright to LGPLv2.1.
  Others will follow.
2009-06-08 22:44:01 +02:00
Andy Wingo
f4aa8d53a0 call-with-values can make fewer closures
* module/language/tree-il.scm: Rename let-exp and letrec-exp to let-body
  and letrec-body. Add <let-values>, a one-expression let-values that
  should avoid the needless creation of two closures in many common
  multiple-value cases. We'll need to add an optimization pass to the
  compiler to produce this form, though, as well as rewriting lambdas
  into lets, etc.

  I added this form instead of adding more special cases to the
  call-with-values compile code because it's a useful intermediate form
  -- it will allow the optimizer to perform constant folding across more
  code.

* module/language/tree-il.scm (parse-tree-il, unparse-tree-il)
  (tree-il->scheme, post-order!, pre-order!): Adapt to let/letrec body
  renaming, and let-values.

* module/language/tree-il/analyze.scm (analyze-lexicals): Adapt for
  renaming, and add cases for let-values.

* module/language/tree-il/compile-glil.scm (flatten): Add a new context,
  `vals', used by let-values code for the values producer. Code that
  produces multiple values can then jump to the let-values MV return
  address directly, instead of trampolining into a procedure. Add code to
  compile let-values.
2009-06-08 00:38:49 +02:00
Andy Wingo
c0ee32452f fix incorrect inlining of + when + is locally redefined
* libguile/vm-i-scheme.c (FUNC2): Use a signed value for the intermediate
  result here. Not sure what the effect is, though.

* module/ice-9/psyntax.scm (chi-top): Toplevel definitions ensure that
  variables are defined in the current module. Fixes the specific case of
  guile-lib's md5.scm, which redefines + -- this code is needed so that
  we don't incorrectly open-code +.

* module/language/tree-il/primitives.scm (resolve-primitives!): I think
  there were some cases in which vars and names would not resolve
  properly here. Fix those.
2009-06-07 00:53:48 +02:00
Andy Wingo
81fd315299 update docs, clean up VM vestiges, macro docs, fix (/ a b c)
* doc/ref/api-procedures.texi (Compiled Procedures): Fix for API changes.

* doc/ref/compiler.texi (Compiling to the Virtual Machine): Replace GHIL
  docs with Tree-IL docs. Update the bits about the Scheme compiler to
  talk about Tree-IL and the expander instead of GHIL. Remove
  <glil-argument>. Add placeholder sections for assembly and bytecode.

* doc/ref/vm.texi: Update examples with what currently happens. Reword
  some things. Fix a couple errors.

* libguile/vm-i-system.c (externals): Remove this instruction, it's not
  used.

* module/ice-9/documentation.scm (object-documentation): If the object is
  a macro, try to return documentation on the macro transformer.

* module/language/assembly/disassemble.scm (disassemble-load-program):
  Fix problem in which we skipped the first element of the object vector,
  because of changes to procedure layouts a few months ago.

* module/language/scheme/spec.scm (read-file): Remove read-file
  definition.

* module/language/tree-il.scm: Reorder exports. Remove <lexical>, it was
  a compat shim to something that was never released. Fix `location'.

* module/language/tree-il/primitives.scm (/): Fix expander for more than
  two args to /.

* module/system/base/compile.scm (read-file-in): Remove unused
  definition.

* module/system/base/language.scm (system): Remove language-read-file.

* module/language/ecmascript/spec.scm (ecmascript): Remove read-file
  definition.
2009-05-24 13:09:01 +02:00
Andy Wingo
55ae815b62 move things to (language tree-il primitives)
* module/language/tree-il/optimize.scm:
* module/language/tree-il/primitives.scm: Move primitive-related things
  to primitive.scm from inline.scm and optimize.scm.

* module/Makefile.am: Update for inventory changes.
2009-05-22 16:07:41 +02:00
Andy Wingo
e0c90f9084 fix tree-il test to work if source info happens to be present
* module/language/tree-il/compile-glil.scm (flatten-lambda): Fix source
  emission.

* test-suite/tests/tree-il.test (strip-source): Strip source info on
  tree-il before compiling, so we don't get extraneous source info in the
  glil. Make check passes!
2009-05-22 13:00:23 +02:00
Andy Wingo
e6b9443179 fix bad call to make-glil-src
* module/language/tree-il/compile-glil.scm (flatten-lambda): Fix bad call
  to make-glil-src, unfortunately not hit during production because
  psyntax doesn't yet understand source locations.
2009-05-22 12:48:45 +02:00
Andy Wingo
0f423f20aa fix apply and call/cc in drop contexts
* module/language/tree-il/compile-glil.scm (flatten): Actually apply only
  needs one arg after the proc. And shit, call/cc and apply in drop
  contexts also need to be able to return arbitrary numbers of values;
  work it by trampolining through their applicative (non-@) definitions.
  Also, simplify the single-valued drop case to avoid the
  truncate-values.

* module/language/tree-il/inline.scm (call/cc):
* module/language/tree-il/optimize.scm (*interesting-primitive-names*):
  Define call/cc as "interesting". Perhaps we should be hashing on value
  and not on variable.

* test-suite/tests/tree-il.test ("application"): Fix up test for new,
  sleeker output. (Actually the GLIL is more verbose, but the assembly is
  better.)
  ("apply", "call/cc"): Add some more tests.
2009-05-21 21:13:24 +02:00
Andy Wingo
30a5e062d0 procedures in "drop" contexts can return unspecified values
* module/language/tree-il/compile-glil.scm (flatten): For applications in
  "drop" context, allow the procedure to return unspecified values
  (including 0 values).

* test-suite/tests/tree-il.test ("application"): Adapt test.

* module/srfi/srfi-18.scm (wrap): Clarify.

* test-suite/tests/srfi-18.test: Fix so that the expression importing
  srfi-18 is expanded before the tests. However the tests are still
  failing, something about 0-valued returns...
2009-05-21 17:22:58 +02:00
Andy Wingo
68623e8e78 remove compile-time-environment
* module/ice-9/boot-9.scm (guile-user): Move the `compile' autoload to
  the guile-user module. Remove reference to compile-time-environment.

* module/language/scheme/compile-ghil.scm:
* module/language/tree-il/compile-glil.scm:
* module/language/tree-il/optimize.scm:
* module/system/base/compile.scm:
* test-suite/tests/compiler.test: Remove definition of and references to
  compile-time-environment. While I do think that recompilation based on
  a lexical environment can be useful, I think it needs to be implemented
  differently. So for now we've lost nothing if we take it away, as it
  doesn't work with syncase anyway.
2009-05-20 17:41:21 +02:00
Andy Wingo
ad9b8c451b fix @slot-ref / @slot-set! compilation
* module/language/tree-il/compile-glil.scm: Add primcall compilers for
  @slot-ref and @slot-set.

* module/language/tree-il/optimize.scm (add-interesting-primitive!): New
  export. Creates an association between a variable in the current module
  and a primitive name.

* module/oop/goops.scm: Rework compiler hooks to work with tree-il and
  not ghil.
2009-05-20 13:59:42 +02:00
Andy Wingo
c11f46afe1 compile list' and vector' to their associated opcodes
* module/language/glil/compile-assembly.scm (glil->assembly): Check the
  length when emitting calls to variable-argument stack instructions.
  Allow two-byte lengths -- allows e.g. calls to `list' with more than
  256 arguments.

* module/language/tree-il/compile-glil.scm: Add primcall associations for
  `list' and `vector', with any number of arguments. Necessary because
  syncase's quasiquote expansions will produce calls to `list' with many
  arguments.

* module/language/tree-il/optimize.scm (*interesting-primitive-names*):
  Add `list' and `vector' to the set of primitives to resolve.
2009-05-20 13:33:44 +02:00
Andy Wingo
5af166bda2 don't allocate too many locals for expansions of `or'
* module/language/tree-il/analyze.scm (analyze-lexicals): Add in a hack
  to avoid allocating more locals than necessary for expansions of `or'.
  Documented in the source.

* test-suite/tests/tree-il.test: Add a test case.
2009-05-20 12:46:23 +02:00
Andy Wingo
e32a1792de a few fixups
* module/ice-9/psyntax.scm (chi-install-global, syntax-case): Fix a
  couple of cases in which bare datums were passed to output
  constructors.

* module/ice-9/psyntax-pp.scm: Regenerated.

* module/language/scheme/spec.scm (scheme): Clean up the #:compilers
  list.

* module/language/tree-il/compile-glil.scm (flatten): Fix call to
  `length' in call/cc compiler.
2009-05-20 11:59:41 +02:00
Andy Wingo
a1a482e0e9 and, or, cond etc use syntax-rules, compile scheme through tree-il
* libguile/vm-i-system.c:
* libguile/vm-engine.h (ASSERT_BOUND): New assertion, that a value is
  bound. Used by local-ref and external-ref in paranoid mode.

* module/ice-9/boot-9.scm (and, or, cond, case, do): Since we are
  switching to use psyntax as the first pass of the compiler, and perhaps
  soon of the interpreter too, we need to make sure it expands out all
  forms to primitive expressions. So define expanders for these derived
  syntax forms, as in the R5RS report.

* module/ice-9/psyntax-pp.scm: Regenerate, with core forms fully
  expanded.

* module/ice-9/psyntax.scm (build-void): New constructor, for making
  undefined values.
  (build-primref): Add in a hack so that primitive refs in the boot
  module expand out to toplevel refs, not module refs.
  (chi-void): Use build-void.
  (if): Define an expander for if that calls build-conditional.

* module/language/scheme/compile-tree-il.scm (compile-tree-il): Use let*
  so as not to depend on binding order for the result of
  (current-module).

* module/language/scheme/spec.scm (scheme): Switch over to tree-il as the
  primary intermediate language. Not yet fully tested, but at least it
  can compile psyntax-pp.scm.

* module/language/tree-il/analyze.scm (analyze-lexicals): Arguments don't
  count towards a function's nlocs.

* module/language/tree-il/compile-glil.scm (*comp-module*, compile-glil):
  Define a "compilation module" fluid.
  (flatten-lambda): Fix a call to make-glil-argument. Fix bug in
  heapifying arguments.
  (flatten): Fix number of arguments passed to apply instruction. Add a
  special case for `(values ...)'. If inlining primitive-refs fails,
  try expanding into toplevel-refs if the comp-module's variable is the
  same as the root variable.

* module/language/tree-il/optimize.scm (resolve-primitives!): Add missing
  src variable for <module-ref>.

* test-suite/tests/tree-il.test ("lambda"): Fix nlocs counts. Add a
  closure test case.
2009-05-20 11:15:22 +02:00
Andy Wingo
ce09ee1989 add tree-il->glil compilation test suite
* module/language/tree-il.scm (parse-tree-il): Fix a number of bugs.
  (unparse-tree-il): Apply takes rest args now.

* module/language/tree-il/analyze.scm (analyze-lexicals)
  (analyze-lexicals): Heap vars shouldn't increment the number of locals.

* module/language/tree-il/optimize.scm (resolve-primitives!): Don't
  resolve public refs to primitives, not at the moment anyway.

* test-suite/Makefile.am (SCM_TESTS): Add tree-il test.

* test-suite/lib.scm (pass-if, expect-fail, pass-if-exception)
  (expect-fail-exception): Rewrite as syntax-rules macros. In a very
  amusing turn of events, it turns out that bindings introduced by
  hygienic macros are not visible inside expansions produced by
  defmacros. This seems to be expected, so go ahead and work around the
  problem.

* test-suite/tests/srfi-31.test ("rec special form"): Expand in eval.

* test-suite/tests/syntax.test ("begin"): Do some more expanding in eval,
  though all is not yet well.

* test-suite/tests/tree-il.test: New test suite, for tree-il->glil
  compilation.
2009-05-18 23:45:35 +02:00
Andy Wingo
dce042f1f7 special cases for more types of known applications
* module/language/tree-il/compile-glil.scm (flatten): Handle a number of
  interesting applications, and fix a bug for calls in `drop' contexts.

* module/language/tree-il/inline.scm: Define expanders for apply,
  call-with-values, call-with-current-continuation, and values.
2009-05-18 01:08:34 +02:00
Andy Wingo
112edbaea3 inline calls to some primitives
* module/system/base/pmatch.scm: Wrap consequents in (let () ) instead of
  (begin ) so that they can have local definitions.

* module/language/tree-il/compile-glil.scm: Inline some calls to
  primitives.
2009-05-17 23:24:26 +02:00
Andy Wingo
1eec95f8de define `delay' in terms of make-promise
* module/ice-9/boot-9.scm (delay): Define `delay' in terms of
  make-promise.

* module/ice-9/psyntax-pp.scm (compile): Regenerated with a fully
  compiled Guile, so that the gensym numbers are the same.

* module/language/tree-il/compile-glil.scm: Add some notes about what
  needs doing to catch up to the old compiler.
2009-05-17 18:04:36 +02:00
Andy Wingo
2ce77f2d95 and now, we residualize the original names into the metadata. yay!
* module/language/tree-il/compile-glil.scm (vars->bind-list)
  (emit-bindings, flatten-lambda, flatten): Write the original names into
  <glil-bind> structures. Yaaaaay!
2009-05-17 16:46:46 +02:00
Andy Wingo
696495f4d2 actually pass original ids on to tree-il data types
* module/ice-9/psyntax.scm (build-lambda, build-let, build-named-let)
  (build-letrec): Actually pass along the original ids to tree-il
  constructors.

* module/ice-9/psyntax-pp.scm: Regenerated.

* module/language/tree-il.scm: Add fields in <lambda>, <let>, and
  <letrec> for the original variable names.

* module/language/tree-il/compile-glil.scm (compile-glil): Adapt for new
  make-lambda arg.
2009-05-17 16:39:55 +02:00
Andy Wingo
cf10678fe7 tree-il -> glil compiler works now, at least in initial tests
* module/language/tree-il/analyze.scm: Break analyzer out into its own
  file.

* module/language/tree-il/compile-glil.scm: Port the GHIL->GLIL compiler
  over to work on tree-il. Works, but still misses a number of important
  optimizations.

* module/language/tree-il.scm: Add <void>. Not used quite yet.

* module/language/glil.scm: Remove <glil-argument>, as it is the same as
  <glil-local> (minus an offset).

* module/language/glil/compile-assembly.scm:
* module/language/glil/decompile-assembly.scm:
* module/language/ghil/compile-glil.scm: Adapt for <glil-argument>
* removal.

* module/Makefile.am (TREE_IL_LANG_SOURCES): Reorder, and add
  analyze.scm.
2009-05-15 23:44:14 +02:00
Andy Wingo
073bb617eb add lexical analyzer and allocator
* module/language/tree-il/optimize.scm: Rework to just export the
  optimize! procedure.

* module/language/tree-il/compile-glil.scm (analyze-lexicals): New
  function, analyzes and allocates lexical variables. Almost ready to
  compile now.
  (codegen): Dedent.
2009-05-14 00:11:54 +02:00
Andy Wingo
cb28c08537 add primitive expander for tree-il
* module/Makefile.am: Add inline.scm.

* module/language/tree-il.scm (pre-order!, post-order!): pre-order! is
  new. post-order! existed but was not public. They do destructive tree
  traversals of tree-il, and need more documentation. Also, add
  predicates to tree-il's export list.

* module/language/tree-il/inline.scm: New file, which expands primitives
  into more primitive primitives. In the future perhaps it will not be
  necessary, as the general inlining infrastructure will handle these
  cases, but for now it's useful.

* module/language/tree-il/optimize.scm: Move post-order! out to better
  pastures.
2009-05-12 22:29:34 +02:00
Andy Wingo
9efc833d65 add tree-il optimizer
* module/language/tree-il/optimize.scm: New module, for optimizations.
  Currently all we have is resolving some toplevel refs to primitive
  refs.

* module/Makefile.am: Add new module.

* module/language/tree-il.scm: Fix exports for accessors for `src'.

* module/language/tree-il/compile-glil.scm: Tweaks, things still aren't
  working yet.
2009-05-11 23:23:34 +02:00
Andy Wingo
b81d329e44 more work on tree-il compilation
* module/language/scheme/amatch.scm: Remove, this approach won't be used.

* module/Makefile.am: Adjust for additions and removals.

* module/language/scheme/compile-ghil.scm: Remove an vestigial debugging
  statement.

* module/language/scheme/spec.scm:
* module/language/scheme/compile-tree-il.scm:
* module/language/scheme/decompile-tree-il.scm: Add tree-il compiler and
  decompiler.

* module/language/tree-il/compile-glil.scm: Add some notes.

* module/language/tree-il/spec.scm: No need to wrap expressions in
  lambdas -- GHIL needs somewhere to put its variables, we don't.
2009-05-08 12:56:18 +02:00
Andy Wingo
811d10f5a2 new language: tree-il. psyntax generates it when run in compile mode.
* module/Makefile.am: Add tree-il sources.

* module/ice-9/compile-psyntax.scm: Adjust for sc-expand producing
  tree-il in compile mode.

* module/ice-9/psyntax.scm: Switch from expand-support to tree-il for
  generating output in compile mode. Completely generate tree-il -- the
  output wasn't Scheme before, but now it's completely not Scheme.

* module/ice-9/psyntax-pp.scm: Regenerated.

* module/language/scheme/compile-ghil.scm: Strip structures using
  tree-il, not expand-support.

* module/language/tree-il.scm:
* module/language/tree-il/spec.scm
* module/language/tree-il/compile-glil.scm: New language. It will compile
  to GLIL, though it doesn't yet.
2009-05-07 17:32:01 +02:00