This suggests moving the conditional that determines if mini-gmp is used
into scmconfig.h.
* configure.ac: replace PKG_CHECK_MODULES for gmp with AC_LIB_HAVE_LINKFLAGS
Remove ENABLE_MINI_GMP define. Also don't run mpz_inits test for
--enable-mini-gmp.
* libguile/gen-scmconfig.c (main) [ENABLE_MINI_GMP]: replace ENABLE_MINI_GMP
with SCM_I_GSC_ENABLE_MINI_GMP
* libguile/bytevectors.c [ENABLE_MINI_GMP]: replace ENABLE_MINI_GMP
with SCM_ENABLE_MINI_GMP
* libguile/init.c [ENABLE_MINI_GMP]: replace ENABLE_MINI_GMP
with SCM_ENABLE_MINI_GMP
* libguile/numbers.c: include scm.h
[SCM_ENABLE_MINI_GMP]: provide mpz_inits and mpz_clears
[ENABLE_MINI_GMP]: prefer SCM_ENABLE_MINI_GMP to ENABLE_MINI_GMP
* libguile/numbers.h: include scm.h
* libguile/random.c [ENABLE_MINI_GMP]: replace ENABLE_MINI_GMP
with SCM_ENABLE_MINI_GMP
* libguile/socket.c [ENABLE_MINI_GMP]: replace ENABLE_MINI_GMP
with SCM_ENABLE_MINI_GMP
* libguile/read.c: Remove support for all non-default reader options.
Also remove support for source positions. The idea is that
primitive-read should just be a stripped-down, easy-to-understand reader
that is enough to bootstrap the C reader. Probably more refactoring
will follow.
* libguile/read.h:
* libguile/read.c (scm_primitive_read): New name for C reader.
(scm_read): Call current value of "read" variable.
(scm_init_read): Initialize "read" binding to "primitive-read".
Replaced later in boot-9.
* module/ice-9/read.scm (read-array): Return pair for dimension when len
is given; single number is lbnd for list->typed-array.
* test-suite/tests/arrays.test: More test cases for the reader.
Instead of defining a separate module, given that "read" calls are quite
all over the place, we're just going to replace the boot "read" binding
with read.scm. This way, we'll be able to remove support for reader
options in the boot reader, as it will only ever be used for a finite
set of files.
* NEWS: Update.
* module/Makefile.am (ice-9/boot-9.go): Depend on read.scm.
(SOURCES):
* am/bootstrap.am (SOURCES): Don't build a ice-9/read.go, as we include
it.
* module/ice-9/boot-9.scm (read-syntax): Define here, as "include" now
uses it.
(read-hash-procedures, read-hash-procedure, read-hash-extend): New
procedures. Will replace C variants.
(read, read-syntax): Include read.scm to define these.
* module/ice-9/psyntax-pp.scm (include): Regenerate.
* module/ice-9/psyntax.scm (include): Use read-syntax, so we get better
source information.
* module/ice-9/read.scm (let*-values): New local definition, to avoid
loading srfi-11.
(%read): Use list->typed-array instead of u8-list->bytevector.
* module/language/scheme/spec.scm: Remove (ice-9 read) import;
read-syntax is there in the boot environment
* libguile/ports.c (peek_iconv_codepoint): If the input has no bytes,
there's little point in raising a decoding error here. Therefore remove
the needless iconv acquisition, harmonising with suspendable-ports.
* test-suite/tests/ports.test ("port-encoding"): Update test to include
some input so that the exception gets raised.
* module/ice-9/psyntax.scm: Use the vector representation of source
properties internally. We have to convert to alists when going to
Tree-IL, but this will be in harmony with syntax objects once the reader
switches to vectors too.
* module/ice-9/psyntax-pp.scm: Regenerate.
* module/system/vm/assembler.scm (intern-constant, link-data): Write the
vector representation of source instead of the alist. Saves a lot of
heap size, object file size, and init time when serializing syntax
objects with source.
* module/system/vm/assembler.scm (<asm>, make-assembler)
(intern-constant, emit-init-constants): Instead of loading a dependent
value each time it's needed in the relocation procedure, eagerly patch
values when they are created. Allows keeping values in registers, which
decreases code size.
* module/ice-9/psyntax.scm: Add a special case for (). There are
already special cases for pairs, vectors, etc; the issue is that with
read-syntax, the () might be come into psyntax as an annotated syntax
object, which here we would want to strip, to preserve the invariant to
psyntax users that all lists are unwrapped.
* module/ice-9/psyntax.scm: With the new behavior of datum->syntax which
allows #f for the lexical context, we have the question of what module
to attach to these newly created syntax objects. In that case we'll
mark down #f as the module, indicating that we know nothing. We have to
extend a number of other cases to default to the expander's idea of the
current module, if a syntax object has no module scope.
Also, change datum->syntax to attach the empty wrap, not the top wrap.
Attaching the top wrap leads to multiply applying the top mark, as you
recurse into subexpressions.
* module/ice-9/psyntax.scm (strip): It used to be that terms in the
source program could have a "top" mark, and when stripping marks we'd
stop recursing when we see an expression with the top mark. This had
the good effect that source programs could contain quoted syntax
objects, or quoted objects with shared structure -- in theory anyway.
In practice the compiler didn't support objects with shared structure.
Anyway when we switch to "read-syntax", quoted expressions can contain
syntax objects introduced by the reader, which naturally we would want
to strip away in a (quote FOO) form. Therefore we remove the
top-marked? optimization.
* module/ice-9/psyntax.scm (quote-syntax): New core form. Usually the
expander will unwrap all syntax objects from the input term. However
sometimes you want to preserve a syntax object, as a datum. That's
when you want quote-syntax.
* module/ice-9/psyntax-pp.scm: Regenerate.
* module/ice-9/local-eval.scm (identifier-syntax-from-box): Use
quote-syntax instead of our datum->syntax trick, which relied on
psyntax's special treatment of the top mark.
* doc/ref/api-macros.texi (Syntax Case): Update documentation for
datum->syntax.
* module/ice-9/psyntax.scm (datum->syntax): Use #:source keyword for
source location info instead of an optional, and allow an alist.
* module/ice-9/psyntax-pp.scm: Regenerate.
* module/ice-9/read.scm (%read, read): Refactor to allow read and
read-syntax to share an implementation.
(read-syntax): New function.
* doc/ref/api-macros.texi (Syntax Case): Document that template-id can
be false, and document srcloc.
* module/ice-9/psyntax-pp.scm: Regenerate.
* module/ice-9/psyntax.scm (syntax?): Allow the lexical context to be
null. Allow srcloc to be a source properties alist. Inspired by
Racket.
* module/ice-9/psyntax.scm (datum->syntax): Add an additional optional
argument, to allow callers to provide source annotation information.
* module/ice-9/psyntax-pp.scm: Regenerate.
* libguile/syntax.c (scm_make_syntax): Add optional "source" argument.
Note that this function is internal.
(scm_syntax_source): New function, replacing definition in boot-9.scm.
* libguile/syntax.h: Add new declarations.
* module/ice-9/psyntax-pp.scm:
* module/ice-9/psyntax.scm (source-annotation): For syntax objects, the
source annotation comes direct from the syntax object.
* module/system/vm/assembler.scm (link-data, intern-constant): Write
5-word syntax objects.
* module/ice-9/read.scm (read): Use R5RS read-char, peek-char,
write-char rather than (ice-9 textual-ports) wrappers, because the
latter are expressed in terms of the former.
