mit-scheme.git
29 years agoAdd conditionalizations for EMX/GCC under OS/2.
Chris Hanson [Mon, 9 Oct 1995 05:56:18 +0000 (05:56 +0000)]
Add conditionalizations for EMX/GCC under OS/2.

29 years agoChanges to get this to work with Visual Age C++.
Chris Hanson [Sun, 8 Oct 1995 15:43:38 +0000 (15:43 +0000)]
Changes to get this to work with Visual Age C++.

29 years agoChange order of #include files, so that "config.h" can override
Chris Hanson [Sun, 8 Oct 1995 15:32:27 +0000 (15:32 +0000)]
Change order of #include files, so that "config.h" can override
default definition of HEAP_FREE in "memmag.h".

29 years agoTest __IBMC__ in addition to __STDC__ because the IBM OS/2 C compiler
Chris Hanson [Sun, 8 Oct 1995 15:31:46 +0000 (15:31 +0000)]
Test __IBMC__ in addition to __STDC__ because the IBM OS/2 C compiler
defines __STDC__ to mean "strict ANSI" rather than "ANSI features
supported".

29 years agoFix bug: define HEAP_FREE for OS/2. The procedure `free' has changed
Chris Hanson [Sun, 8 Oct 1995 15:28:45 +0000 (15:28 +0000)]
Fix bug: define HEAP_FREE for OS/2.  The procedure `free' has changed
in the most recent release of the IBM C compiler, and barfs when it is
handed a null pointer.  Previously, it did nothing -- but it had not
been my intention to take advantage of this behavior.

29 years agoTest __IBMC__ in addition to __STDC__ because the IBM OS/2 C compiler
Chris Hanson [Sun, 8 Oct 1995 15:23:41 +0000 (15:23 +0000)]
Test __IBMC__ in addition to __STDC__ because the IBM OS/2 C compiler
defines __STDC__ to mean "strict ANSI" rather than "ANSI features
supported".

29 years agoChange order of #include files, so that "config.h" can override
Chris Hanson [Sun, 8 Oct 1995 15:22:15 +0000 (15:22 +0000)]
Change order of #include files, so that "config.h" can override
default definition of HEAP_FREE in "memmag.h".

29 years agoDired mode should not be case sensitive; this was changed between
Chris Hanson [Fri, 6 Oct 1995 21:01:06 +0000 (21:01 +0000)]
Dired mode should not be case sensitive; this was changed between
Emacs 18 and 19 and the case-insensitive default makes more sense.

29 years agoRetract the most recent changes to get Linux ELF binaries to work.
Chris Hanson [Fri, 6 Oct 1995 06:45:24 +0000 (06:45 +0000)]
Retract the most recent changes to get Linux ELF binaries to work.
These changes do not and cannot work, because signal handlers cannot
be run while the special Scheme segment registers are in use.  The
reason is that the signal handlers are registered as 32-bit offsets,
and the associated segment descriptors are not registered with those
offsets; when the signal arrives, the offset is used with whatever
segment descriptor happens to be in the CS register.  When the Scheme
segment descriptors are in place, this offset is meaningless and
causes Scheme to die immediately with a SIGSEGV.

One possible workaround for this problem would be to re-register the
signal handlers whenever the segment registers are hacked, but this is
very expensive in time, and the Scheme<->C interface is already too
slow.  The only other feasibility is to get the Linux kernel changed
so that signal handlers are invoked in a known segment without regard
to the contents of CS at the time of the signal; but although this is
a sensible idea, it's necessary to convince the kernel developer's of
it's desirability, and meanwhile the existing Linux installations will
not work.

29 years agoTweaks to previous change.
Chris Hanson [Fri, 6 Oct 1995 06:27:44 +0000 (06:27 +0000)]
Tweaks to previous change.

29 years agoChange conditionalization for Linux ELF so that it is set in a single
Chris Hanson [Thu, 5 Oct 1995 06:33:15 +0000 (06:33 +0000)]
Change conditionalization for Linux ELF so that it is set in a single
place.  This simplifies building a non-ELF binary if desired.

29 years agoImplement changes needed for Linux ELF binary format. This format,
Chris Hanson [Thu, 5 Oct 1995 03:34:50 +0000 (03:34 +0000)]
Implement changes needed for Linux ELF binary format.  This format,
unlike the older a.out format, translates the data segment to begin at
0x08000000, and additionally changes the calling conventions so that
returned structures are allocated by the caller and passed in as
pointers.  We fix the first problem by extending the win32s segment
register manipulation to also be used with Linux ELF.  The second
problem is fixed by extending the OS/2 assembly-language patches to
also cover Linux ELF.

29 years agoThe Free register can't be compared to the value in REGBLOCK_MEMTOP
Chris Hanson [Thu, 5 Oct 1995 03:27:42 +0000 (03:27 +0000)]
The Free register can't be compared to the value in REGBLOCK_MEMTOP
unless it is translated from a "C" address to a "Scheme" address.
This makes a difference on the i386 where we sometimes use the segment
registers in a way that makes these two representations different.

29 years agoMerge 8.0 and 7.4 versions of this file.
Chris Hanson [Wed, 4 Oct 1995 22:53:58 +0000 (22:53 +0000)]
Merge 8.0 and 7.4 versions of this file.

29 years agoChange compressed-file support to match changes to unix support.
Chris Hanson [Tue, 3 Oct 1995 21:14:24 +0000 (21:14 +0000)]
Change compressed-file support to match changes to unix support.

29 years agoMake sure that the value from GROUP-INSERT-FILE! is returned by
Chris Hanson [Tue, 3 Oct 1995 21:12:37 +0000 (21:12 +0000)]
Make sure that the value from GROUP-INSERT-FILE! is returned by
%INSERT-FILE.

29 years agoChange encoded file I/O methods to generate messages to indicate their
Chris Hanson [Tue, 3 Oct 1995 19:15:54 +0000 (19:15 +0000)]
Change encoded file I/O methods to generate messages to indicate their
activity.  Change READ-COMPRESSED-FILE to uncompress the file to a
temporary file, then read the temporary file into the buffer, as this
makes it possible to read larger files without exhausting the heap.

29 years agoDon't generate file read/write messages when using special file I/O
Chris Hanson [Tue, 3 Oct 1995 19:01:01 +0000 (19:01 +0000)]
Don't generate file read/write messages when using special file I/O
methods.  It is expected that these methods will have appropriate
messages of their own.

29 years agoPatches to get the microcode to compile properly on Linux ELF
Chris Hanson [Sun, 1 Oct 1995 07:19:42 +0000 (07:19 +0000)]
Patches to get the microcode to compile properly on Linux ELF
systems.

29 years agoChanged time again.
Stephen Adams [Fri, 29 Sep 1995 20:25:04 +0000 (20:25 +0000)]
Changed time again.

29 years agoIncreased benchmark time.
Stephen Adams [Fri, 29 Sep 1995 20:20:31 +0000 (20:20 +0000)]
Increased benchmark time.

29 years agoPerformed by SRA: Add some declarations to increase performance.
Chris Hanson [Fri, 29 Sep 1995 19:57:46 +0000 (19:57 +0000)]
Performed by SRA: Add some declarations to increase performance.

29 years agoInitial revision
Stephen Adams [Fri, 29 Sep 1995 17:50:10 +0000 (17:50 +0000)]
Initial revision

29 years agoRemoved flatten.
Stephen Adams [Fri, 29 Sep 1995 17:05:50 +0000 (17:05 +0000)]
Removed flatten.

29 years agoChange code to look for encoded files or ".info" suffix. Add switch
Chris Hanson [Thu, 28 Sep 1995 16:17:13 +0000 (16:17 +0000)]
Change code to look for encoded files or ".info" suffix.  Add switch
to disable selection highlighting (GJS likes selections but not
highlighting).

29 years agoAdd extra argument to GET-PATHNAME-OR-ALTERNATE, to allow it to be
Chris Hanson [Thu, 28 Sep 1995 16:11:30 +0000 (16:11 +0000)]
Add extra argument to GET-PATHNAME-OR-ALTERNATE, to allow it to be
used in place of FILE-EXISTS?.

29 years agoChange interrupt locking of COMPUTE-KEY-HASH to avoid turning off the
Chris Hanson [Thu, 28 Sep 1995 06:18:47 +0000 (06:18 +0000)]
Change interrupt locking of COMPUTE-KEY-HASH to avoid turning off the
GC interrupt.  This was causing fatal lossage for BDC.

29 years agoApparently the unix definition of completion-ignored-extensions was
Chris Hanson [Thu, 28 Sep 1995 06:08:16 +0000 (06:08 +0000)]
Apparently the unix definition of completion-ignored-extensions was
missing all of the standard Scheme extensions.

29 years agoImplement X selection mechanism for doing cut and paste between Edwin
Chris Hanson [Wed, 27 Sep 1995 16:24:28 +0000 (16:24 +0000)]
Implement X selection mechanism for doing cut and paste between Edwin
and other X windows.  This change requires recent changes to the
microcode.

There appears to be a bug in the implementation of the INCR protocol
for receiving large selections -- after working on this for a day, I'm
giving up and leaving it broken, since it will probably never be
used.

29 years agoMore changes for selection support. These changes are required for
Chris Hanson [Wed, 27 Sep 1995 16:21:46 +0000 (16:21 +0000)]
More changes for selection support.  These changes are required for
the current Edwin code.

29 years agoFixed bug with windows closed by user between last GC and exiting.
Stephen Adams [Mon, 25 Sep 1995 20:54:10 +0000 (20:54 +0000)]
Fixed bug with windows closed by user between last GC and exiting.

29 years agoHandle symbolic link listings specially -- they require extra parsing
Chris Hanson [Mon, 25 Sep 1995 20:29:00 +0000 (20:29 +0000)]
Handle symbolic link listings specially -- they require extra parsing
to extract the file name.

29 years agoAdd primitive to perform XSelectInput library call.
Chris Hanson [Mon, 25 Sep 1995 20:04:19 +0000 (20:04 +0000)]
Add primitive to perform XSelectInput library call.

29 years agoAdded win32 to list.
Stephen Adams [Mon, 25 Sep 1995 18:12:05 +0000 (18:12 +0000)]
Added win32 to list.

29 years agoPut in MIT copyright notice.
Stephen Adams [Mon, 25 Sep 1995 18:11:12 +0000 (18:11 +0000)]
Put in MIT copyright notice.

29 years agoFixed bug in DeleteDIB.
Stephen Adams [Mon, 25 Sep 1995 18:10:46 +0000 (18:10 +0000)]
Fixed bug in DeleteDIB.
Rewrote loop to speed up DIBSetPixelsUnaligned by 2x-3x.
Put in MIT copyright notice.

29 years agoImplement some new primitives to support X selections.
Chris Hanson [Mon, 18 Sep 1995 22:52:36 +0000 (22:52 +0000)]
Implement some new primitives to support X selections.

29 years agoImplement some new primitives to support X selections.
Chris Hanson [Mon, 18 Sep 1995 22:49:07 +0000 (22:49 +0000)]
Implement some new primitives to support X selections.

29 years agoAdd mechanism to support conversion of primitive arguments from Scheme
Chris Hanson [Mon, 18 Sep 1995 22:33:38 +0000 (22:33 +0000)]
Add mechanism to support conversion of primitive arguments from Scheme
integers to C `unsigned long' integers.

29 years agoMoved benchmarks source directory.
Stephen Adams [Sat, 16 Sep 1995 21:02:35 +0000 (21:02 +0000)]
Moved benchmarks source directory.

29 years ago*** empty log message ***
Stephen Adams [Sat, 16 Sep 1995 20:46:20 +0000 (20:46 +0000)]
*** empty log message ***

29 years agoAdd option to ignore the button events associated with a click that
Chris Hanson [Fri, 15 Sep 1995 19:28:51 +0000 (19:28 +0000)]
Add option to ignore the button events associated with a click that
selects the window.

29 years agoPrevious attempt to implement a mechanism to restore the signal mask
Chris Hanson [Fri, 15 Sep 1995 05:51:12 +0000 (05:51 +0000)]
Previous attempt to implement a mechanism to restore the signal mask
when aborting to the interpreter was flawed: the signal mask was saved
once, restored on the first abort, and left unchanged after that.
The flaw is repaired by re-saving the signal mask after every abort.

29 years agoInitial revision
Stephen Adams [Thu, 14 Sep 1995 13:22:12 +0000 (13:22 +0000)]
Initial revision

29 years agoChanges dealing with encoded files:
Chris Hanson [Wed, 13 Sep 1995 23:01:05 +0000 (23:01 +0000)]
Changes dealing with encoded files:

1. Appending to an encoded file works by reading the file into a
   temporary buffer, appending the text, and writing it back out.

2. Line translation is handled for encoded files the same way it is
   for unencoded files.

3. Doing I/O to non-encoded files, if the file doesn't exist, an
   encoded file is used in its place if available.

29 years agoChange SUBPROCESS-I/O-PORT to allow setting of input and output line
Chris Hanson [Wed, 13 Sep 1995 21:25:13 +0000 (21:25 +0000)]
Change SUBPROCESS-I/O-PORT to allow setting of input and output line
translation independently.

29 years agoChange SUBPROCESS-I/O-PORT to allow setting of input and output line
Chris Hanson [Wed, 13 Sep 1995 19:58:39 +0000 (19:58 +0000)]
Change SUBPROCESS-I/O-PORT to allow setting of input and output line
translation independently.

29 years agoChange MAKE-GENERIC-I/O-PORT to allow setting of input and output line
Chris Hanson [Wed, 13 Sep 1995 19:56:13 +0000 (19:56 +0000)]
Change MAKE-GENERIC-I/O-PORT to allow setting of input and output line
translation independently.

29 years agoAlways write RMAIL and unix mail files with line-translation disabled.
Chris Hanson [Wed, 13 Sep 1995 04:28:04 +0000 (04:28 +0000)]
Always write RMAIL and unix mail files with line-translation disabled.
Scheme will read such files fine, and this simplifies sharing of the
files between operating systems.

29 years agoAlways write RMAIL and unix mail files with line-translation disabled.
Chris Hanson [Wed, 13 Sep 1995 04:18:04 +0000 (04:18 +0000)]
Always write RMAIL and unix mail files with line-translation disabled.
Scheme will read such files fine, and this simplifies sharing of the
files between operating systems.

29 years agoReplace Scheme variables *TRANSLATE-FILE-DATA-ON-INPUT?* and
Chris Hanson [Wed, 13 Sep 1995 03:57:22 +0000 (03:57 +0000)]
Replace Scheme variables *TRANSLATE-FILE-DATA-ON-INPUT?* and
*TRANSLATE-FILE-DATA-ON-OUTPUT?* with corresponding editor variables.
This allows the variables to be overridden on a per-buffer basis.

29 years agoFix bug in previous change.
Chris Hanson [Mon, 11 Sep 1995 22:43:49 +0000 (22:43 +0000)]
Fix bug in previous change.

29 years agoAdd new condition type ILLEGAL-PATHNAME-COMPONENT so that pathname
Chris Hanson [Mon, 11 Sep 1995 21:25:45 +0000 (21:25 +0000)]
Add new condition type ILLEGAL-PATHNAME-COMPONENT so that pathname
parsing errors can be caught.

29 years agoAdd error-control to the prompt completion code. If any of the
Chris Hanson [Mon, 11 Sep 1995 21:24:41 +0000 (21:24 +0000)]
Add error-control to the prompt completion code.  If any of the
completion operations signals an error, the error will be caught and
the user informed that the typein string is in error.  This is
particularly useful for such things as pathnames, which have a fairly
restricted syntax.

29 years agoAdd new condition type ILLEGAL-PATHNAME-COMPONENT so that pathname
Chris Hanson [Mon, 11 Sep 1995 21:03:08 +0000 (21:03 +0000)]
Add new condition type ILLEGAL-PATHNAME-COMPONENT so that pathname
parsing errors can be caught.

29 years agoAdd new condition type ILLEGAL-PATHNAME-COMPONENT so that pathname
Chris Hanson [Mon, 11 Sep 1995 20:52:26 +0000 (20:52 +0000)]
Add new condition type ILLEGAL-PATHNAME-COMPONENT so that pathname
parsing errors can be caught.

29 years agoAdd new condition type ILLEGAL-PATHNAME-COMPONENT so that pathname
Chris Hanson [Mon, 11 Sep 1995 19:07:16 +0000 (19:07 +0000)]
Add new condition type ILLEGAL-PATHNAME-COMPONENT so that pathname
parsing errors can be caught.

29 years agoAdded ABS.
Stephen Adams [Mon, 11 Sep 1995 14:29:12 +0000 (14:29 +0000)]
Added ABS.

29 years agoAdded methods for {CEILING,FLOOR,ROUND,TRUNCATE}->EXACT.
Stephen Adams [Mon, 11 Sep 1995 14:26:45 +0000 (14:26 +0000)]
Added methods for {CEILING,FLOOR,ROUND,TRUNCATE}->EXACT.

29 years agoAdded {CEILING,FLOOR,ROUND,TRUNCATE}->EXACT to global references.
Stephen Adams [Mon, 11 Sep 1995 13:35:59 +0000 (13:35 +0000)]
Added {CEILING,FLOOR,ROUND,TRUNCATE}->EXACT to global references.

29 years agoKludged out the interrupt checks in continuations from primitive
Stephen Adams [Fri, 8 Sep 1995 16:49:08 +0000 (16:49 +0000)]
Kludged out the interrupt checks in continuations from primitive
calls: unlike a continuation from calling arbitrary code, the
primitive can't be the procedure we are in, so if this path is on a
loop, the loop is broken elsewhere (in the procedure or continuation
that calls the primitive.)

29 years agoAdded LIST-REF to global expansions.
Stephen Adams [Fri, 8 Sep 1995 03:09:28 +0000 (03:09 +0000)]
Added LIST-REF to global expansions.

29 years agoAdded expansion for global LIST-REF.
Stephen Adams [Fri, 8 Sep 1995 03:09:09 +0000 (03:09 +0000)]
Added expansion for global LIST-REF.

29 years agoFixed dbg-info for diamonds.
Stephen Adams [Fri, 8 Sep 1995 03:07:41 +0000 (03:07 +0000)]
Fixed dbg-info for diamonds.

29 years agoFixed bug with top-level forms wit internal continuations.
Stephen Adams [Fri, 8 Sep 1995 02:56:57 +0000 (02:56 +0000)]
Fixed bug with top-level forms wit internal continuations.

29 years agoRemoved dead code.
Stephen Adams [Fri, 8 Sep 1995 02:44:17 +0000 (02:44 +0000)]
Removed dead code.

29 years agoAdded FORM/EXACT-INTEGER?
Stephen Adams [Fri, 8 Sep 1995 00:56:01 +0000 (00:56 +0000)]
Added FORM/EXACT-INTEGER?

29 years agoRepaginate.
Stephen Adams [Thu, 7 Sep 1995 22:07:49 +0000 (22:07 +0000)]
Repaginate.

29 years agoAdded more global expansions.
Stephen Adams [Thu, 7 Sep 1995 20:53:56 +0000 (20:53 +0000)]
Added more global expansions.

29 years agoCommented out WIDEN.
Stephen Adams [Tue, 5 Sep 1995 21:45:17 +0000 (21:45 +0000)]
Commented out WIDEN.
It is failing in the pre-cps position.
Test case: dragon4.scm:dragon-fixup

29 years agoFixed call to ERROR:WRONG-TYPE-DATUM to pass correct number of
Stephen Adams [Tue, 5 Sep 1995 21:27:35 +0000 (21:27 +0000)]
Fixed call to ERROR:WRONG-TYPE-DATUM to pass correct number of
arguments.

29 years agoFix typo.
Stephen Adams [Tue, 5 Sep 1995 19:45:32 +0000 (19:45 +0000)]
Fix typo.

29 years agoAdded constant-folder for STRING->SYMBOL.
Stephen Adams [Tue, 5 Sep 1995 19:14:35 +0000 (19:14 +0000)]
Added constant-folder for STRING->SYMBOL.

29 years agoInitial revision
Stephen Adams [Tue, 5 Sep 1995 19:08:51 +0000 (19:08 +0000)]
Initial revision

29 years agoMoved MONOTONIC-STRONG-EQ-HASH-TABLEs to base/fasthash.scm
Stephen Adams [Tue, 5 Sep 1995 19:08:20 +0000 (19:08 +0000)]
Moved MONOTONIC-STRONG-EQ-HASH-TABLEs to base/fasthash.scm

29 years agoAdded more types.
Stephen Adams [Tue, 5 Sep 1995 19:07:29 +0000 (19:07 +0000)]
Added more types.
Added support for abstractly testing (OBJECT-TYPE? <number> ...)
Removed all the construction of a database of types for known
operators.  That is the province of typedb.scm (and perhaps
fakeprim.scm)

29 years agoAdded STRING->SYMBOL.
Stephen Adams [Tue, 5 Sep 1995 19:04:43 +0000 (19:04 +0000)]
Added STRING->SYMBOL.

29 years agoFixed problem with barfing at %make-heap-closure in predicate position
Stephen Adams [Tue, 5 Sep 1995 19:04:13 +0000 (19:04 +0000)]
Fixed problem with barfing at %make-heap-closure in predicate position
(how it got there is another mystery).

Replaced many primitive procedures by their `%' variants.

29 years agoMoved split and widen pre-CPS.
Stephen Adams [Tue, 5 Sep 1995 19:01:45 +0000 (19:01 +0000)]
Moved split and widen pre-CPS.

29 years agoAdded `ignored' bindings to keep SF quiet.
Stephen Adams [Tue, 5 Sep 1995 19:00:21 +0000 (19:00 +0000)]
Added `ignored' bindings to keep SF quiet.

29 years agoRemoved some primitives (CAR, STRING-SET!, ...) from the simple
Stephen Adams [Tue, 5 Sep 1995 18:59:19 +0000 (18:59 +0000)]
Removed some primitives (CAR, STRING-SET!, ...) from the simple
operator list.  Added an set of replacement `%primitives' as unchecked
replacements that do no type or range checking.  The ordinary
primitives are now coded as calls to the primitive procedure and may
be used to raise errors.

29 years agoRestructured generic arithetic rewrites. Removed possibility of doing
Stephen Adams [Tue, 5 Sep 1995 18:56:00 +0000 (18:56 +0000)]
Restructured generic arithetic rewrites.  Removed possibility of doing
early `diamond' rewrites - that is the province of typerew and laterew.

Removed kludged type-checked stuff into typerew.

29 years agoFixed errant paren.
Stephen Adams [Tue, 5 Sep 1995 18:48:37 +0000 (18:48 +0000)]
Fixed errant paren.

29 years agoRearranged files in midend.
Stephen Adams [Tue, 5 Sep 1995 18:46:39 +0000 (18:46 +0000)]
Rearranged files in midend.

29 years agoFirst `working' version.
Stephen Adams [Tue, 5 Sep 1995 18:46:09 +0000 (18:46 +0000)]
First `working' version.

29 years agoChanged `dethunking' to expect promises, so we can profile primitive
Stephen Adams [Tue, 5 Sep 1995 18:01:24 +0000 (18:01 +0000)]
Changed `dethunking' to expect promises, so we can profile primitive
procedures.

29 years agoTafting
Stephen Adams [Tue, 5 Sep 1995 12:31:38 +0000 (12:31 +0000)]
Tafting

29 years agoMoved FOR-EVERY to utils.scm
Stephen Adams [Mon, 4 Sep 1995 21:55:10 +0000 (21:55 +0000)]
Moved FOR-EVERY to utils.scm

29 years agoInitial revision
Stephen Adams [Mon, 4 Sep 1995 21:07:36 +0000 (21:07 +0000)]
Initial revision

29 years agoPunted reference to obsolete %vector-index.
Stephen Adams [Mon, 4 Sep 1995 20:41:14 +0000 (20:41 +0000)]
Punted reference to obsolete %vector-index.

29 years agoAdded an efficiency tip.
Stephen Adams [Mon, 4 Sep 1995 14:00:54 +0000 (14:00 +0000)]
Added an efficiency tip.

29 years agoSafety checkin.
Stephen Adams [Sun, 3 Sep 1995 17:15:04 +0000 (17:15 +0000)]
Safety checkin.

29 years agoAdded expansions for %generic-index-check/ref and %generic-index-check/set!.
Stephen Adams [Sat, 2 Sep 1995 20:57:24 +0000 (20:57 +0000)]
Added expansions for %generic-index-check/ref and %generic-index-check/set!.

29 years agoCheck in prior to rewrite to split type analysis from rewrites.
Stephen Adams [Sat, 2 Sep 1995 13:30:23 +0000 (13:30 +0000)]
Check in prior to rewrite to split type analysis from rewrites.

29 years agoFixed bug whereby primitives that were called with the wrong number of
Stephen Adams [Fri, 1 Sep 1995 21:35:49 +0000 (21:35 +0000)]
Fixed bug whereby primitives that were called with the wrong number of
arguments were invoked directly.  Now they are applied, to produce an
error.

29 years agoInitial revision
Stephen Adams [Fri, 1 Sep 1995 18:53:45 +0000 (18:53 +0000)]
Initial revision

29 years agoExcised %machine-fixnum?
Stephen Adams [Thu, 31 Aug 1995 15:26:05 +0000 (15:26 +0000)]
Excised %machine-fixnum?

29 years agoRemoved "Node absent" warning.
Stephen Adams [Wed, 30 Aug 1995 19:54:46 +0000 (19:54 +0000)]
Removed "Node absent" warning.

29 years agoAdded more global operators.
Stephen Adams [Tue, 29 Aug 1995 14:56:39 +0000 (14:56 +0000)]
Added more global operators.

29 years agoAdded feature where (ACCESS <name> <system-global-environment>) is
Stephen Adams [Tue, 29 Aug 1995 14:06:45 +0000 (14:06 +0000)]
Added feature where (ACCESS <name> <system-global-environment>) is
unsyntaxed as <name> if UNSYNTAXER:ELIDE-GLOBAL-ACCESSES? is non-false
and there are no shadowing bindings.

This is slightly imperfect because the unsyntaxer cant know about
names bound in the scode that surrounds the expression that was
initially passed to UNSYNTAX.  Perhaps these names should be a
parameter.  Debuggers could then augment the bound names with names
from the environment (or perhaps unsyntaxing should be done with
reference to an environment).

The feature is currently OFF by default, but I envision it being
necessary for sanity as more and more IEEE/R4RS operators are
integrated as global accesses (so that the compiler knows that it can
optimize them).