Followup to 58f86505d6 ("Augment
`Libs.private' with for non-GNU platforms.").
Reported by Eli Zaretskii <eliz@gnu.org>.
* meta/guile-2.0-uninstalled.pc.in (Libs): Add all the optional
libraries, as done in commit 58f86505d6.
* test-suite/lib.scm (clear-stale-stack-references): New helper.
* test-suite/tests/gc.test:
* test-suite/tests/threads.test: Use the new helper.
I removed the "unresolved" throw, as I'd like to see what the bots do
with this.
Version 2.0.4 was pushed with the version info of 2.0.3, which is
incorrect. Version 2.0.5 fixes that.
* GUILE-VERSION (GUILE_MICRO_VERSION): Increment.
(LIBGUILE_INTERFACE_CURRENT): Increment to account for new C functions
`scm_c_value_ref', `scm_call', etc.
(LIBGUILE_INTERFACE_AGE): Increment.
* module/ice-9/local-eval.scm (local-wrap): Fix the (module? e) case, to
reference the expression 'x' instead of the non-existent variable
'exp', as was previously done. Also use quasisyntax instead of
quasiquote, so that the introduced 'lambda' is an identifier instead
of a bare symbol, so that this will work in modules that have rebound
'lambda' to something else.
* test-suite/tests/eval.test (local-eval): Make sure to test both
'local-eval' and 'local-compile' when the specified environment is a
module.
Reported by Eli Zaretskii <eliz@gnu.org>.
* meta/guile-2.0.pc.in (Libs.private): Add all the optional libraries
that may be pulled via Gnulib on non-GNU platforms.
* libguile/i18n.c (scm_make_locale): Null out c_locale_name after it's
freed, so it doesn't get freed again. While this only has effect if
USE_GNU_LOCALE_API is unset, the nulling out happens in both cases,
to be consistent and robust.
(scm_nl_langinfo): Free c_result in all switch cases, in particular
P_SIGN_POSN, N_SIGN_POSN, INT_P_SIGN_POSN, and INT_N_SIGN_POSN.
* libguile/vm-engine.h (DEAD): New macro, nulls out a value.
* libguile/vm-i-system.c:
* libguile/vm-i-loader.c:
* libguile/vm-i-scheme.c: Use DEAD when variables become dead.
Later we can #ifdef this out, but I want to give the buildbots a try
with this patch to make sure it's correct.
* module/ice-9/psyntax.scm (syntax-type): Return an additional value
that contains the entire form in _all_ cases, including for definition
forms. Previously, the entire form was not returned for definition
forms.
(expand-expr): Add an additional argument that contains the entire
form in _all_ cases, including for definition forms. Use it to
include the entire form in error messages, notably for definitions in
expression context. Include the source location information, which
was previously missing from these errors when the rhs expression was
an atom. Improve the "definition in expression context" error message
to be more comprehensible for Scheme beginners.
(expand-top-sequence, expand, expand-body): Adjust as needed to handle
the additional return value from 'syntax-type' and the additional
argument to 'expand-expr'.
* module/ice-9/psyntax-pp.scm: Regenerate.
* NEWS: Update.
* module/language/tree-il/peval.scm (peval):
* module/language/tree-il/primitives.scm (dynamic-wind): When you make a
gensym that just has to be compared against other gensyms, it will be
unique if the prefix doesn't end in something that can be interpreted
as a number. There's no reason to make that character something
difficult like " ". So change to use a dash in that case.
* module/ice-9/psyntax-pp.scm: Regenerate. More readable now.
* libguile/macros.c (scm_init_macros): Add definition of new
syntax-session-id helper.
* module/ice-9/psyntax.scm: Capture a reference to syntax-session-id at
boot time. Uniquify marks and labels using the session id.
* module/ice-9/boot-9.scm: Shunt syntax-session-id off to (system
syntax) once we finished booting.
* module/ice-9/compile-psyntax.scm: Override syntax-session-id when
remaking psyntax to avoid spurious diffs.
* module/ice-9/psyntax-pp.scm: Regenerate.
This reverts commit ad432bc831.
Not all gensyms need to be universally-unique: most of them just need to
be unique within some portion of a Guile session. We'll take a
different tack on this problem in the next commit.
* module/ice-9/local-eval.scm: New module (ice-9 local-eval) which
exports `the-environment', `local-eval', and `local-compile'.
* libguile/debug.c (scm_local_eval): New C function that calls the
Scheme implementation of `local-eval' in (ice-9 local-eval).
* libguile/debug.h (scm_local_eval): Add prototype.
* doc/ref/api-evaluation.texi (Local Evaluation): Add documentation.
* test-suite/tests/eval.test (local evaluation): Add tests.
* test-suite/standalone/test-loose-ends.c (test_scm_local_eval):
Add test.
* module/Makefile.am: Add ice-9/local-eval.scm.
Based on a patch by Mark H Weaver <mhw@netris.org>.
* module/language/tree-il/analyze.scm
(format-analysis)[check-simple-format-args]: New procedure. Use it.
Add support for applications of <module-ref>.
* module/system/base/message.scm (%warning-types): Handle the `format
simple-format' warning.
* module/language/scheme/spec.scm (scheme)[make-default-environment]:
Use `simple-format' as the default `format'.
* test-suite/tests/tree-il.test ("warnings")["format"]: Explicitly use
(@ (ice-9 format) format) where needed.
("simple-format"): New test prefix.
* module/ice-9/boot-9.scm:
* module/ice-9/psyntax.scm (syntax-module, syntax-local-binding)
(syntax-locally-bound-identifiers): After boot, move these definitions
to a new (system syntax) module.
* module/ice-9/psyntax-pp.scm: Regenerate.
* doc/ref/api-macros.texi: Add some words about syntax-module and
friends being in (system syntax).
* test-suite/tests/gc.test ("gc"): Fix "lexical vars are collectable"
test. (Previously, the binding was getting inlined, so we weren't
testing what we meant to test. Besides that, the value was a
constant, not a closure, so it wasn't collectable in the first place!)
* module/language/tree-il/peval.scm (peval): Fold (values
'singly-valued-expression) to 'singly-valued-expression in contexts
that expect multiple values, in addition to those that expect single
values.
* module/language/tree-il/compile-glil.scm (flatten-lambda-case): Fix
miscompilation of `values' in a push context with RA.
* test-suite/tests/tree-il.test: Add low-level test for this
miscompilation.
Fixes part of <http://bugs.gnu.org/10520>.
Reported by Bruno Haible <bruno@clisp.org>.
* libguile/vm-i-system.c (COMPILER_BARRIER): New macro.
(halt, return, return_values): Insert compiler barrier after each
assignment to `fp'.
This is a followup to 690a0112e5 ("Remove
the "has finalizer?" bit from pointer objects.")
* libguile/foreign.c (scm_set_pointer_finalizer_x): Leave the type cell
unchanged. Before, `equal?' would break on pointers on which
`set-pointer-finalizer!' had been called.
* test-suite/tests/foreign.test ("make-pointer")["equal? modulo
finalizer (set-pointer-finalizer!)"]: New test.
* module/ice-9/boot-9.scm (current-filename): Canonicalize the path, so
that the result is independent of the current directory, and so
that `dirname' can traverse up the file system, as in the
add-to-load-path example.
(add-to-path): Remove. The eval-when semantics make this macro too
tricky to explain -- people will start using it on other path-like
things, in lexical contours, whereas it only really makes sense with
load-like paths at the toplevel.
* doc/ref/api-evaluation.texi (Loading): Fix link to load-with-path
docs, and remove add-to-path docs.
* libguile/hashtab.c (scm_i_hashtable_print): Use `SCM_UNPACK', not
`SCM2PTR'.
* test-suite/tests/hash.test: Adjust tests that rely on the output of
the printer to just check for the suffix of the hash table's external
representation.
