sbcl.git
15 years ago1.0.25.2: Eliminate untagged pointers to heap space in cold-init
Alastair Bridgewater [Tue, 3 Feb 2009 04:13:13 +0000 (04:13 +0000)]
1.0.25.2: Eliminate untagged pointers to heap space in cold-init

Load-time value fixups encountered by genesis are added as a kind of
toplevel form to be patched during cold-init.  The reference to the
location to fix up was being dumped as a SAP pointing to the correct
point in heap space.  Instead of dumping a SAP, we now dump the
containing object and an offset within the object, thus removing one
obstacle to running the GC or doing other heap-space relocation prior
to running the cold-toplevels.

15 years ago1.0.25.1: x86-64 code fixup recording for gc / slash-and-burn
Alastair Bridgewater [Tue, 3 Feb 2009 04:11:05 +0000 (04:11 +0000)]
1.0.25.1: x86-64 code fixup recording for gc / slash-and-burn

x86-64 code segments do not have absolute references to within
themselves, nor do they have relative references to without themselves,
making them relocatable without patching.  The GC has long since been
updated to reflect this, but the fixup recording code originally part
of the x86 port had been retained.

  * Removed x86-64 code-object fixup recording code everywhere.

  * Added some commentary to x86iod fixup handling in genesis.

15 years ago1.0.25: release, will be tagged as sbcl_1_0_25
Juho Snellman [Tue, 3 Feb 2009 02:49:12 +0000 (02:49 +0000)]
1.0.25: release, will be tagged as sbcl_1_0_25

15 years ago1.0.24.48: Do explicit sign-extension of small signed alien return values
Juho Snellman [Mon, 26 Jan 2009 01:54:47 +0000 (01:54 +0000)]
1.0.24.48: Do explicit sign-extension of small signed alien return values

        * gcc 4.3 has a different interpretation of the ABI, and
          zero-extends signed chars, shorts and (on x86-64) ints.

15 years ago1.0.24.47: Fix 1.0.21.29 regression with enums in structs.
Christophe Rhodes [Wed, 21 Jan 2009 17:29:08 +0000 (17:29 +0000)]
1.0.24.47: Fix 1.0.21.29 regression with enums in structs.

This is not the most elegant of fixes, but arrange to return the old
structure where applicable, and make incompatible mentions cause a
CERROR with clobber continue semantics, to parallel what I think is the
logic in the union/struct case.

15 years ago1.0.24.46: SB-BSD-SOCKETS workingness from saved cores on Windows
Nikodemus Siivola [Sat, 17 Jan 2009 10:56:47 +0000 (10:56 +0000)]
1.0.24.46: SB-BSD-SOCKETS workingness from saved cores on Windows

 * Need to call WSA-STARTUP on init.

 * Reported by Stephen Westfold, fix by Rudi Schlatte.

15 years ago1.0.24.45: handle IO errors in LDB and when saving the core
Nikodemus Siivola [Sat, 17 Jan 2009 10:43:57 +0000 (10:43 +0000)]
1.0.24.45: handle IO errors in LDB and when saving the core

 * In LDB, if users response cannot be read, assume y.

 * Check fwrite() called when saving the core, and report the error.

 * Patch by Daniel Lowe.

15 years ago1.0.24.44: bug in INVALID-ARRAY-INDEX-ERROR, leaving EXPECTED-TYPE slot unbound
Nikodemus Siivola [Fri, 16 Jan 2009 12:13:40 +0000 (12:13 +0000)]
1.0.24.44: bug in INVALID-ARRAY-INDEX-ERROR, leaving EXPECTED-TYPE slot unbound

 * It's :EXPECTED-TYPE, not :TYPE. ...having compiler check calls to ERROR like
   this would be nice.

 * Marginally better test to catch this.

15 years ago1.0.24.43: DEFTYPE to accept bodies consisting a single symbol
Nikodemus Siivola [Thu, 15 Jan 2009 09:19:31 +0000 (09:19 +0000)]
1.0.24.43: DEFTYPE to accept bodies consisting a single symbol

 * Regression from 1.0.22.8, reported by Ariel Badichi.

15 years ago1.0.24.42: fix bug 235a
Nikodemus Siivola [Wed, 14 Jan 2009 18:37:20 +0000 (18:37 +0000)]
1.0.24.42: fix bug 235a

 AKA https://bugs.launchpad.net/sbcl/+bug/309141

 * Replace DEFINED-FUN-FUNCTIONAL with DEFINED-FUN-FUNCTIONALS, and reuse
   the functional only if policy matches.

15 years ago1.0.24.41: add necessary disambiguating curly braces to runtime
Nikodemus Siivola [Mon, 12 Jan 2009 16:52:19 +0000 (16:52 +0000)]
1.0.24.41: add necessary disambiguating curly braces to runtime

 * When QSHOW is undefined FSHOW(...) expands into nothing, leading to unexpected
   results for code like

     if (gencgc_verbose)
       FSHOW(...)

   as the conditional will then apply to the next syntactic element.

 * Patch by Daniel Lowe.

15 years ago1.0.24.40: fix regression from 1.0.24.37
Nikodemus Siivola [Mon, 12 Jan 2009 16:20:37 +0000 (16:20 +0000)]
1.0.24.40: fix regression from 1.0.24.37

 * TNs with KIND :CONSTANT don't have leaves if they represent load
   time values.

 * Reported by Attila Lendvai.

15 years ago1.0.24.39: mutex changes
Gabor Melis [Mon, 12 Jan 2009 15:00:21 +0000 (15:00 +0000)]
1.0.24.39: mutex changes

- do what a FIXME suggests and rename MUTEX-VALUE to MUTEX-OWNER
- in the process, make sure that the value returned is less stale
- keep MUTEX-VALUE around for compatibility for a while
- also add HOLDING-MUTEX-P
- to make MUTEX-OWNER and HOLDING-MUTEX-P useful make unithread builds
  keep track of the owner of mutex

15 years ago1.0.24.38: improve test scripts
Gabor Melis [Mon, 12 Jan 2009 14:57:08 +0000 (14:57 +0000)]
1.0.24.38: improve test scripts

- robustify threads.test.sh even more
- fix the same bug related to $! in finalize.test.sh
- make condition-wait-sigcont.lisp work unithread builds

15 years ago1.0.24.37: full raw-instance-slot support on HPPA
Nikodemus Siivola [Mon, 12 Jan 2009 11:47:57 +0000 (11:47 +0000)]
1.0.24.37: full raw-instance-slot support on HPPA

 * Patch by Larry Valkama.

15 years ago1.0.24.36: don't print the array object when reporting index out of bounds
Nikodemus Siivola [Mon, 12 Jan 2009 11:26:42 +0000 (11:26 +0000)]
1.0.24.36: don't print the array object when reporting index out of bounds

 * Report the type of the array instead, and encapsulate the array object in the
   condition object so that it can still be inspected manually.

 * Suggested by Stas Boukarev.

15 years ago1.0.24.35: Flag-setting VOPs on x86[-64] and conditional moves
Paul Khuong [Sun, 11 Jan 2009 18:39:07 +0000 (18:39 +0000)]
1.0.24.35: Flag-setting VOPs on x86[-64] and conditional moves

 * Most :CONDITIONAL VOPs only specify which condition flags they set

 * GENERIC-{EQL,=,<,>} are :CONDITIONAL VOPs, but don't show up as
   calls anymore

 * Values may be selected with CMOVcc if applicable (and :CMOV is
   in *backend-subfeatures*, for x86):
    - Values that are represented, unboxed, in GPRs are CMOVed using
      custom VOPs
    - Unboxed float and complex types aren't converted
    - Other types are assumed to be boxed and CMOVed as descriptors

 * A test to try and to cover an interesting cross-section of flags
   and values to move conditionally.

15 years ago1.0.24.34: IR2: additional representation for predicates, conditional moves
Paul Khuong [Sun, 11 Jan 2009 18:33:31 +0000 (18:33 +0000)]
1.0.24.34: IR2: additional representation for predicates, conditional moves

 * :CONDITIONAL VOPs can now specify how to interpret the test
   they compute without performing the branch directly. How the
   test is specified is completely platform-dependent and only
   affects new-style :CONDITIONAL VOPs and a new BRANCH-IF VOP
   (src/compiler/$ARCH/pred.lisp).

 * Candidates for conversion to conditional moves are found
   and may be converted, depending on CONVERT-CONDITIONAL-MOVE-P,
   a new VM support routine. C-C-M-P returns NIL to punt on the
   conversion, or 5 values:
    1. name of the VOP to use
    2. TN for the first argument (NIL if none)
    3. TN for the second argument (NIL if none)
    4. TN for the result
    5. A list of info data, which will be appended to the flags

   The correct values will be MOVEd in the argument TNs if needed
   before computing the condition, and the result MOVEd to the right
   TN after the conditional move VOP.

15 years ago1.0.24.33: fix bug 316075
Nikodemus Siivola [Sun, 11 Jan 2009 16:34:00 +0000 (16:34 +0000)]
1.0.24.33: fix bug 316075

 https://bugs.launchpad.net/sbcl/+bug/316075

 * In DEFINE-ALIEN-ROUTINE, if the return-type is VOID, don't return
   the result of ALIEN-FUNCALL but a literal NIL instead.

15 years ago1.0.24.32: undo parts of 1.0.24.26
Gabor Melis [Sun, 11 Jan 2009 15:56:03 +0000 (15:56 +0000)]
1.0.24.32: undo parts of 1.0.24.26

No need for memory barriers when unlocking a spinlock on x86/x86-64.
The ordering rules and the cache coherency mechanism together
guarantee this. However, the compiler must be prevented from
reordering instructions with the unlock (at least in one direction).
This is now done in the runtime, but not in Lisp as the Lisp compiler
does no reordering.

15 years ago1.0.24.31: robustify threads.test.sh some more
Nikodemus Siivola [Sun, 11 Jan 2009 15:38:46 +0000 (15:38 +0000)]
1.0.24.31: robustify threads.test.sh some more

 * Missed --no-userinit --no-sysinit --disable-debugger, causing it to
   fail for some setups.

15 years ago1.0.24.30: fixed and tested some more cleanups on hppa-hpux
Nikodemus Siivola [Sat, 10 Jan 2009 11:19:22 +0000 (11:19 +0000)]
1.0.24.30: fixed and tested some more cleanups on hppa-hpux

 * Fix a stray #+ -> #!+.

 * Removed unneeded nops.

 * Explanation of magic numbers (but not yet substituted.)

   (Above changes in patch by Larry Valkama)

 * Fix a bunch of comments in the HPPA backend to use the right number
   of semicolons, and use FIXME-lav instead of FIX-lav to mark things
   (better grepping for the rest of us.)

15 years ago1.0.24.29: fix a typo in documentation
Nikodemus Siivola [Sat, 10 Jan 2009 10:51:13 +0000 (10:51 +0000)]
1.0.24.29: fix a typo in documentation

15 years ago1.0.24.28: make unbind in the runtime zero the value of the binding
Gabor Melis [Fri, 9 Jan 2009 17:47:47 +0000 (17:47 +0000)]
1.0.24.28: make unbind in the runtime zero the value of the binding

... to parallel what the Lisp side does. As far as I see the rationale
for doing that (in doc/internals/specials.texinfo) does not apply here
since signals are blocked when the dynbind functions are called. That
said, instead of adding documentation let's fix it.

15 years ago1.0.24.27: target-thread cosmetics
Gabor Melis [Fri, 9 Jan 2009 16:45:17 +0000 (16:45 +0000)]
1.0.24.27: target-thread cosmetics

- in the docstring of CONDITION-NOTIFY document that the mutex must be held
- remove and explain fixme in CONDITION-WAIT
- respect 80 char limit
- use ; ;; ;;; and ;;;; where appropriate
- fill the comment paragraphs
- add form feeds (^L) to separate pages

15 years ago1.0.24.26: fix release spinlock
Gabor Melis [Fri, 9 Jan 2009 16:43:56 +0000 (16:43 +0000)]
1.0.24.26: fix release spinlock

Both in the runtime and in Lisp releasing a spinlock was a simple
assignment. That doesn't work because the new value may not make it to
main memory by the time another CPU wants to acquire it making it
needlessly slow. Worse, it also allows the CPU to reorder instructions
from the critical section after the release.

There is a spinlock implementation for MIPS in the runtime, but it's
not used as we don't have threads on that platform. I don't know if
it's broken too.

15 years ago1.0.24.25: add volatile after asm in spinlock and swap_lispobjs
Gabor Melis [Fri, 9 Jan 2009 16:42:09 +0000 (16:42 +0000)]
1.0.24.25: add volatile after asm in spinlock and swap_lispobjs

... to prevent the compiler from optimizing away certain calls or
maybe reorder them. Test for swap_lispobj.

15 years ago1.0.24.24: fix threads.test.sh
Gabor Melis [Thu, 8 Jan 2009 17:21:08 +0000 (17:21 +0000)]
1.0.24.24: fix threads.test.sh

... so that it doesn't leave stray sbcl child processes around and it
actually tests what it wants to test.

15 years ago1.0.24.23: fix alien struct struct member offset bug
Christophe Rhodes [Mon, 5 Jan 2009 08:55:20 +0000 (08:55 +0000)]
1.0.24.23: fix alien struct struct member offset bug

No-one uses struct struct members, right?  Well, academics are
notoriously bad at keeping up to dat with good practice...

15 years ago1.0.24.22: mudball of VOP updates for HPPA
Nikodemus Siivola [Sun, 4 Jan 2009 07:49:02 +0000 (07:49 +0000)]
1.0.24.22: mudball of VOP updates for HPPA

 * Based on a mix of the old hppa-code and the mips backend.

 * Patch by Larry Valkama.

15 years ago1.0.24.21: call stub needed to switch between hpux heap-spaces
Nikodemus Siivola [Sun, 4 Jan 2009 07:35:53 +0000 (07:35 +0000)]
1.0.24.21: call stub needed to switch between hpux heap-spaces

 * Patch by Larry Valkama.

15 years ago1.0.24.20: misc HPPA & HPUX updates
Nikodemus Siivola [Sun, 4 Jan 2009 07:23:03 +0000 (07:23 +0000)]
1.0.24.20: misc HPPA & HPUX updates

 * "To make it compile and run".

 * Config.hppa-hpux missed from 1.0.24.18.

 * Patch by Larry Valkama.

15 years ago1.0.24.19: COMPILE-TIME reports timings at millisecond accuracy
Nikodemus Siivola [Sat, 3 Jan 2009 17:05:44 +0000 (17:05 +0000)]
1.0.24.19: COMPILE-TIME reports timings at millisecond accuracy

 * Patch by Luis Oliveira.

15 years ago1.0.24.18: new HPUX specific files
Nikodemus Siivola [Sat, 3 Jan 2009 16:26:22 +0000 (16:26 +0000)]
1.0.24.18: new HPUX specific files

 * Also more separation of linux stuff versus common stuff (hpux vs linux).

 * Patch by Larry Valkama.

15 years ago1.0.24.17: grab-bag of fixes to make hpux-os smile
Nikodemus Siivola [Sat, 3 Jan 2009 16:17:48 +0000 (16:17 +0000)]
1.0.24.17: grab-bag of fixes to make hpux-os smile

 * Patch by Larry Valkama.

15 years ago1.0.24.16: updates on how we deal with fixup on HPPA
Nikodemus Siivola [Sat, 3 Jan 2009 16:14:03 +0000 (16:14 +0000)]
1.0.24.16: updates on how we deal with fixup on HPPA

 * Patch by Larry Valkama.

15 years ago1.0.24.15: contrib fixes for HPPA
Nikodemus Siivola [Sat, 3 Jan 2009 16:10:19 +0000 (16:10 +0000)]
1.0.24.15: contrib fixes for HPPA

 * Fix or disable what breaks in contribs, so not everything breaks.

 * Patch by Larry Valkama.

15 years ago1.0.24.14: fix what seems to be namespace collision by HPUX headers
Nikodemus Siivola [Sat, 3 Jan 2009 16:02:31 +0000 (16:02 +0000)]
1.0.24.14: fix what seems to be namespace collision by HPUX headers

 * Patch by Larry Valkama.

15 years ago1.0.24.13: solve overlapping mmap and munmap slices on HPUX
Nikodemus Siivola [Sat, 3 Jan 2009 15:59:44 +0000 (15:59 +0000)]
1.0.24.13: solve overlapping mmap and munmap slices on HPUX

 * Patch by Larry Valkama.

15 years ago1.0.24.12: adding and fixing the HPUX/HPPA build target
Nikodemus Siivola [Sat, 3 Jan 2009 15:50:46 +0000 (15:50 +0000)]
1.0.24.12: adding and fixing the HPUX/HPPA build target

 * Patch by Larry Valkama.

15 years ago1.0.24.11: stack allocation support for HPPA
Nikodemus Siivola [Sat, 3 Jan 2009 15:41:58 +0000 (15:41 +0000)]
1.0.24.11: stack allocation support for HPPA

 * Thiemo Seufer's MIPS stack allocation work and other things by him
   ported over to HPPA.

 * Patch by Larry Valkama.

15 years ago1.0.24.10: raw slot support for HPPA
Nikodemus Siivola [Sat, 3 Jan 2009 15:39:38 +0000 (15:39 +0000)]
1.0.24.10: raw slot support for HPPA

 * Remove raw slot support workaround on hppa, VOPs implemented
   instead.

 * Patch by Larry Valkama.

15 years ago1.0.24.9: fix overlapping address spaces on sparc
Nikodemus Siivola [Sat, 3 Jan 2009 15:17:44 +0000 (15:17 +0000)]
1.0.24.9: fix overlapping address spaces on sparc

 * Patch by Bruce O'Neel.

15 years ago1.0.24.8: fix scav_lose
Gabor Melis [Fri, 2 Jan 2009 18:37:55 +0000 (18:37 +0000)]
1.0.24.8: fix scav_lose

Used to do one pointer dereferencing too many when printing the
widetag of the lispobj leading to memory faults, ultimately obscuring
the backtrace instead of losing cleanly. Memory faults during gc could
cause all kinds of trouble rendering the backtrace even less
informative.

15 years ago1.0.24.7: CHAR-CODE type derivation
Nikodemus Siivola [Fri, 2 Jan 2009 16:05:28 +0000 (16:05 +0000)]
1.0.24.7: CHAR-CODE type derivation

 * Patch by Paul Khuong.

15 years ago1.0.24.6: OPTIMIZATION #23 is there already
Nikodemus Siivola [Fri, 2 Jan 2009 15:11:08 +0000 (15:11 +0000)]
1.0.24.6: OPTIMIZATION #23 is there already

 * Type information for &REST lists is available after constraint
   propagation.

15 years ago1.0.24.5: SB-INTROSPECT: rename FUNCTION-ARGLIST to FUNCTION-LAMBDA-LIST
Nikodemus Siivola [Fri, 2 Jan 2009 14:24:03 +0000 (14:24 +0000)]
1.0.24.5: SB-INTROSPECT: rename FUNCTION-ARGLIST to FUNCTION-LAMBDA-LIST

 * Deprecate FUNCTION-ARGLIST.

 * Improve the docstring.

 * Original patch by Tobias Rittweiler.

15 years ago1.0.24.4: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST
Nikodemus Siivola [Fri, 2 Jan 2009 13:55:36 +0000 (13:55 +0000)]
1.0.24.4: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST

 * Patch by Tobias Rittweiler.

15 years ago1.0.24.3: sanity check address spaces
Nikodemus Siivola [Fri, 2 Jan 2009 12:42:08 +0000 (12:42 +0000)]
1.0.24.3: sanity check address spaces

 * Genesis to check that spaces don't overlap.

 * At startup make sure --dynamic-space-size doesn't overflow, or run
   into any space possibly on top of dynamic space. (GENCGC only.)

15 years ago1.0.24.2: CONSTANTP aware GET-SETF-EXPANDER
Nikodemus Siivola [Fri, 2 Jan 2009 11:14:27 +0000 (11:14 +0000)]
1.0.24.2: CONSTANTP aware GET-SETF-EXPANDER

 * Or rather GET-SETF-METHOD-INVERSE -- check for constant arguments,
   which don't need to be rebound. This allows compiler macros for
   SETF-functions to see their constant arguments.

 * This exposes a small thinko in ACCESSOR-VALUES-INTERNAL (something
   gets optimized during PCL build which wasn't before): EARLY-P there
   doesn't mean the method is early.

15 years ago1.0.24.1: Reading from a TWO-WAY-STREAM need not flush the output stream.
Gabor Melis [Thu, 1 Jan 2009 20:50:41 +0000 (20:50 +0000)]
1.0.24.1: Reading from a TWO-WAY-STREAM need not flush the output stream.

In fact, it unnecessarily complicates writing thread-safe code as
readers and writers access the output stream concurrently.

15 years ago1.0.24: release, will be tagged as sbcl_1_0_24
Christophe Rhodes [Wed, 31 Dec 2008 13:27:21 +0000 (13:27 +0000)]
1.0.24: release, will be tagged as sbcl_1_0_24

Also, include build fix for sparc (Juho Snellman / Bruce O'Neel
sbcl-devel 30-12-2008) and mark test :throw :no-such-tag as failing on
x86/linux and x86-64/linux (because they do for me).

15 years ago1.0.23.72: missing NEWS entries for this month
Nikodemus Siivola [Wed, 31 Dec 2008 11:44:37 +0000 (11:44 +0000)]
1.0.23.72: missing NEWS entries for this month

 * Also add fill-column to modeline.

15 years ago1.0.23.71: fix overlapping spaces on OpenBSD
Nikodemus Siivola [Wed, 31 Dec 2008 09:51:47 +0000 (09:51 +0000)]
1.0.23.71: fix overlapping spaces on OpenBSD

 * Patch by Josh Elsasser. Going is despite the freeze since
   overlapping spaces are just plain broken.

 * Also mark THROW NO-SUCH-TAG failing on OpenBSD.

15 years ago1.0.23.70: Add a keyword to DIRECTORY to suppress symlink resolution.
Richard M Kreuter [Fri, 26 Dec 2008 18:33:56 +0000 (18:33 +0000)]
1.0.23.70: Add a keyword to DIRECTORY to suppress symlink resolution.

* Contributed by TC-Rucho.

15 years ago1.0.23.69: Add docstrings to SB-BSD-SOCKETS:SOCKET-MAKE-STREAM.
Richard M Kreuter [Fri, 26 Dec 2008 15:43:25 +0000 (15:43 +0000)]
1.0.23.69: Add docstrings to SB-BSD-SOCKETS:SOCKET-MAKE-STREAM.

* Contributed by Robert Goldman.

15 years ago1.0.23.68: Cleanups in constraint propagation.
Richard M Kreuter [Fri, 26 Dec 2008 14:19:06 +0000 (14:19 +0000)]
1.0.23.68: Cleanups in constraint propagation.

* Three changes here:

(1) Have the conset's min slot always be a fixnum.  The min and max
    slots should now conform to CL sequence bounding index idioms.

(2) Update the extrema in parallel, rather than in sequence, in the
    conset-union, -intersection, -difference.

(3) Remove some noise from conset-intersection that probably included
    an off-by-one error.

* Fixes a bug reported by Tobias C. Rittweiler on sbcl-devel.

15 years ago1.0.23.67: Fix MIPS FIXED-ALLOC VOP.
Thiemo Seufer [Thu, 25 Dec 2008 20:36:19 +0000 (20:36 +0000)]
1.0.23.67: Fix MIPS FIXED-ALLOC VOP.

  * No observable change in behaviour. Apparently nothing ever allocates
    a non-pointer referenced object via this code path.

15 years ago1.0.23.66: Calculate array sizes in a more reliable way.
Thiemo Seufer [Tue, 23 Dec 2008 14:10:23 +0000 (14:10 +0000)]
1.0.23.66: Calculate array sizes in a more reliable way.

  * The old implementation depended on the array header size being
    an even number of words.

  * Also, another micro-optimization for MIPS.

15 years ago1.0.23.65: MIPS runtime micro-optimization.
Thiemo Seufer [Tue, 23 Dec 2008 00:52:13 +0000 (00:52 +0000)]
1.0.23.65: MIPS runtime micro-optimization.

15 years ago1.0.23.64: fixed bug 395
Nikodemus Siivola [Mon, 22 Dec 2008 13:38:04 +0000 (13:38 +0000)]
1.0.23.64: fixed bug 395

 * Add support for base-strings in fill-pointer output streams.

 * Also fix a bug revealed by this change in derivation of
   ARRAY-ELEMENT-TYPE return type.

15 years ago1.0.23.63: WITH-HASH-TABLE-ITERATOR lambda-list beautification
Nikodemus Siivola [Mon, 22 Dec 2008 12:00:24 +0000 (12:00 +0000)]
1.0.23.63: WITH-HASH-TABLE-ITERATOR lambda-list beautification

 * Iterator lambda-list name is NAME, not FUNCTION, which was a bit
   confusing since it is bound to a macro.

 * Thanks to Tobias Rittweiler.

15 years ago1.0.23.62: fix bug 357
Nikodemus Siivola [Mon, 22 Dec 2008 10:50:35 +0000 (10:50 +0000)]
1.0.23.62: fix bug 357

 Originally reported by Bruno Haible, more recently by Stephen Wilson.

 * SHARED-INITIALIZE (SLOT-OBJECT) should not check structure slots
   versus +SLOT-UNBOUND+: uninitialized slots are zeroed. Since adding
   slots to structure classes cannot cause those slots to be added to
   structure instances, we don't really have to check for boundness at
   all.

 * SB-PCL::STRUCTURE-TYPE-SLOT-DESCRIPTION-LIST and
   SB-PCL::MAKE-STRUCTURE-CLASS-DEFSTRUCT-FORM did not take overridden
   slot specifications into account, and the latter also omitted
   initform and type information.

 * Delete SB-PCL::ALLOCATE-STRUCTURE-INSTANCE, unused.

 * ALLOCATE-INSTANCE (STRUCTURE-OBJECT) should not fall back on
   ALLOCATE-STANDARD-INSTANCE.

15 years ago1.0.23.62: Micro-optimization for MIPS' SYMBOL-HASH VOP.
Thiemo Seufer [Sun, 21 Dec 2008 17:55:18 +0000 (17:55 +0000)]
1.0.23.62: Micro-optimization for MIPS' SYMBOL-HASH VOP.

  * MIPS sucks now less.

15 years ago1.0.23.61: tweak concurrent hash-table access check
Nikodemus Siivola [Sun, 21 Dec 2008 15:03:11 +0000 (15:03 +0000)]
1.0.23.61: tweak concurrent hash-table access check

 * Moves the with-concurrent-access-check inside the body where
   the lock is already held. Patch by Attila Lendvai.

 * Fix various attributions in NEWS, caught by TCR.

15 years ago1.0.23.60: fix bug 354: XEPs in backtraces, properly this time
Nikodemus Siivola [Sun, 21 Dec 2008 09:51:01 +0000 (09:51 +0000)]
1.0.23.60: fix bug 354: XEPs in backtraces, properly this time

 * Don't terminate the block in MAYBE-TERMINATE-BLOCK even if the node
   is a call to a function that never returns if it is also the tail
   end of a XEP -- this allows TCO to deal with the XEP.

 * More stale bugs:
   ** 143 -- cannot replicate, interrupt handling has been robustified
      and partially redesigned since than, so confidence that this is
      really gone is reasonably high.
   ** 238 -- has gotten fixed at some point.

15 years ago1.0.23.59: bug 3b has been fixed a while now
Nikodemus Siivola [Fri, 19 Dec 2008 19:00:01 +0000 (19:00 +0000)]
1.0.23.59: bug 3b has been fixed a while now

 * Test case added.

15 years ago1.0.23.58: bug 405 has been fixed a while now
Nikodemus Siivola [Fri, 19 Dec 2008 18:03:09 +0000 (18:03 +0000)]
1.0.23.58: bug 405 has been fixed a while now

 * Fixed as of 1.0.19.32.

15 years ago1.0.23.57: rewrite bug 217
Nikodemus Siivola [Fri, 19 Dec 2008 16:14:58 +0000 (16:14 +0000)]
1.0.23.57: rewrite bug 217

 * Case A is a non-issue. SBCL interprets "unpredictable but harmless"
   to mean "anything, but heap will no be corrupted, data lost, etc"
   -- that is, an error may or may not be signalled, but it the lisp
   session will remain sane and operable.

 * Case B is negated by the portability argument: since specifying both
   is unspecified, we should not build useful-seeming extensions on top
   of it, as it will only create portability problems for users.

 * Case C remains partially valid, except for the runtime WARNINGs.
   Integrated into the new text.

15 years ago1.0.23.56: special variables cause special cases in CLOS cleverness
Nikodemus Siivola [Fri, 19 Dec 2008 15:20:43 +0000 (15:20 +0000)]
1.0.23.56: special variables cause special cases in CLOS cleverness

 * As we cannot reliably ensure nothing calls SET or (SETF
   SYMBOL-VALUE) on the special variable, we need to disable
   optimizations perutation vector optimizations for them.

 * For the same reason we cannot implicitly declare types for even
   local specials in DEFMETHODs.

 * Delete bug 276: the issue it refers to was fixed back when we
   stopped inserting declarations for special variables.

15 years ago1.0.23.55: three stale bugs
Nikodemus Siivola [Fri, 19 Dec 2008 13:46:34 +0000 (13:46 +0000)]
1.0.23.55: three stale bugs

 * 332, 369, 380: added test-cases which pass.

15 years ago1.0.23.54: compiler-macros for WRITE and WRITE-TO-STRING
Nikodemus Siivola [Thu, 18 Dec 2008 20:41:57 +0000 (20:41 +0000)]
1.0.23.54: compiler-macros for WRITE and WRITE-TO-STRING

 * The common use-cases have only constant keywords, but due to the
   way they are defined inlining doesn't really help -- so do it with
   compiler-macros, which bind only those specials specified in the
   call.

15 years ago1.0.23.53: FORMAT performance tweaking 2
Nikodemus Siivola [Thu, 18 Dec 2008 20:39:38 +0000 (20:39 +0000)]
1.0.23.53: FORMAT performance tweaking 2

 * Rearrange FORMAT-FIXED and open code guts of FORMAT-FIXED-AUX for
   both single and double-float cases, gaining ~5% speedup for ~F.

15 years ago1.0.23.52: FORMAT performance tweaking
Nikodemus Siivola [Thu, 18 Dec 2008 11:56:47 +0000 (11:56 +0000)]
1.0.23.52: FORMAT performance tweaking

 * Handle plain ~D using explicitly bindings and OUTPUT-OBJECT to
   avoid paying for WRITE keyword argument parsing.

 * Compile format control strings when SPEED = SPACE.

 * Always transform FORMAT calls when the second argument is a
   function -- trying to save space there doesn't make much sense.

15 years ago1.0.23.51: wrong return value from READ-SEQUENCE :START on file streams
Nikodemus Siivola [Thu, 18 Dec 2008 10:48:21 +0000 (10:48 +0000)]
1.0.23.51: wrong return value from READ-SEQUENCE :START on file streams

 * Regression since 1.0.12.22.

 * Reported by Daniel Herring, patch by Paul Huong.

 * Also tweak the "write a line with bignum characters test" to run a bit faster.

15 years ago1.0.23.50: Less magic constants in the MIPS backend.
Thiemo Seufer [Thu, 18 Dec 2008 00:53:15 +0000 (00:53 +0000)]
1.0.23.50: Less magic constants in the MIPS backend.

15 years ago1.0.23.49: Eliminate FCN as function moniker.
Thiemo Seufer [Wed, 17 Dec 2008 20:36:18 +0000 (20:36 +0000)]
1.0.23.49: Eliminate FCN as function moniker.

15 years ago1.0.23.48: REPL recovers from *READ-SUPPRESS* T with a warning
Nikodemus Siivola [Wed, 17 Dec 2008 17:34:17 +0000 (17:34 +0000)]
1.0.23.48: REPL recovers from *READ-SUPPRESS* T with a warning

 * Reported by Daniel Herring.

15 years ago1.0.23.47: binaries built on now Leopard run on Tiger as well
Nikodemus Siivola [Wed, 17 Dec 2008 17:12:57 +0000 (17:12 +0000)]
1.0.23.47: binaries built on now Leopard run on Tiger as well

 * -mmacosx-version-min=10.4 needed in LINKFLAGS as well.

15 years ago1.0.23.46: Remove redundant uses of zero-tn in the MIPS backend.
Thiemo Seufer [Wed, 17 Dec 2008 14:43:26 +0000 (14:43 +0000)]
1.0.23.46: Remove redundant uses of zero-tn in the MIPS backend.

15 years ago1.0.23.45: Remove superfluous register :offsets in the MIPS backend.
Thiemo Seufer [Wed, 17 Dec 2008 14:42:15 +0000 (14:42 +0000)]
1.0.23.45: Remove superfluous register :offsets in the MIPS backend.

15 years ago1.0.23.44: Nicer code for MIPS, lifted from the SPARC backend.
Thiemo Seufer [Wed, 17 Dec 2008 14:39:28 +0000 (14:39 +0000)]
1.0.23.44: Nicer code for MIPS, lifted from the SPARC backend.

  * Add some :note annotations.

  * Kill a few magic constants.

15 years ago1.0.23.43: More consistent coding style for mips/array.lisp
Thiemo Seufer [Wed, 17 Dec 2008 14:36:11 +0000 (14:36 +0000)]
1.0.23.43: More consistent coding style for mips/array.lisp

15 years ago1.0.23.42: don't grab *WORLD-LOCK* in CLASSOID-TYPEP
Nikodemus Siivola [Tue, 16 Dec 2008 16:37:07 +0000 (16:37 +0000)]
1.0.23.42: don't grab *WORLD-LOCK* in CLASSOID-TYPEP

 * A shoddy bandaid to make Slime less prone to deadlock against the
   compiler when using :SPAWN.

 * Proper fix is to be much, much more principled about grabbing the
   *WORLD-LOCK* in the compiler: we should only ever grab it in places
   where we cannot end up calling arbitary user code while holding
   it.

15 years ago1.0.23.41: fix DX-COMBINATION-P
Nikodemus Siivola [Mon, 15 Dec 2008 09:40:07 +0000 (09:40 +0000)]
1.0.23.41: fix DX-COMBINATION-P

 * Punt on multiple result-uses. Reported by Pascal Costanza and
   Thiemo Seufer.

 * Also check for clean arg flow all cases.

15 years ago1.0.23.40: export page sizes to C with LU suffix
Nikodemus Siivola [Sat, 13 Dec 2008 10:52:07 +0000 (10:52 +0000)]
1.0.23.40: export page sizes to C with LU suffix

 * Rename GENCGC-PAGE-SIZE and *BACKEND-PAGE-SIZE* to
   GENCGC-PAGE-BYTES and *BACKEND-PAGE-BYTES* respectively.

 * Clean up constant.h generation: instead of guessing when to add an
   U suffix, specify when the value is "large", and then add an LU
   suffix.

 * Without the LU suffix some C compilers chose to truncate results of
   some operations where these quantities featured, leading at least
   to an upper limit of #xffff0000 bytes in dynamic space on certain
   64 bit systems.

15 years ago1.0.23.39: Red Hat Enterprise 3 mmap randomization workaround
Nikodemus Siivola [Fri, 12 Dec 2008 15:51:28 +0000 (15:51 +0000)]
1.0.23.39: Red Hat Enterprise 3 mmap randomization workaround

 * Patch by Thomas Burdick: RHEL 3 suffers from the randomized mmap
   nonesense, and reports kernel version 2.4.21. With the attached
   patch, it SBCL runs correctly on these machines.

15 years ago1.0.23.38: fix bug 430 (stack alloc by nested defstruct constructors)
Nikodemus Siivola [Fri, 12 Dec 2008 13:05:23 +0000 (13:05 +0000)]
1.0.23.38: fix bug 430 (stack alloc by nested defstruct constructors)

 * Mark lambdas introduced by the compiler as such, so that
   LAMBDA-SYSTEM-LAMBDA-P returns true for them.

 * Allow USE-GOOD-FOR-DX-P to inspect COMBINATIONs with CLAMBDA
   functionals: if the return value of the function always originates
   from a known DX-capable combination, and the arguments of the
   original combination are used only by the DX-capable combination,
   consider the original combination good for DX.

 * Allow USE-GOOD-FOR-DX-P to inspect REFs to LAMBDA-VARs: if the var
   is bound by a system lambda, has no other refs, is never set, gets
   its value from a single-value combination, and the LVAR it gets its
   value from is good for DX ... then the REF is good for DX as well.

 * HANDLE-NESTED-DYNAMIC-EXTENT-LVARS handles REFs as well by
   recursing on the lvar the REF gets its value from.

15 years ago1.0.23.37: more CLOS and classoid thread safety
Nikodemus Siivola [Fri, 12 Dec 2008 12:27:00 +0000 (12:27 +0000)]
1.0.23.37: more CLOS and classoid thread safety

 * Rename *BIG-COMPILER-LOCK* as *WORLD-LOCK*.

 * Use it to protect classoids, layouts/wrappers, obsolete instance
   updating, etc.

 * Rename sevaral functions which assume their callers are holding the
   lock to have % prefix. Perhaps we should have a separate prefix
   convention for "requires called to lock"? Or a nifty macro layer?
   (Actually there are probably places where read/write locks (or
   STM!)  would be a win...)

 * ENSURE-CLASS-USING-CLASS needs to set up type translations only
   while PCL is being build.

 * DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to be thread
   safe.

 * Test-case for parallel defclass and make-instance.

15 years ago1.0.23.36: typecheck :ALLOCATION :CLASS slot initforms in safe code
Nikodemus Siivola [Fri, 12 Dec 2008 10:57:52 +0000 (10:57 +0000)]
1.0.23.36: typecheck :ALLOCATION :CLASS slot initforms in safe code

 * Initforms for shared slots are not applied at make-instance, but at
   class definition time. (See CLHS 4.3.6 and 7.1.) Reported by Didier
   Verna.

15 years ago1.0.23.35: CLOS tweaking
Nikodemus Siivola [Fri, 12 Dec 2008 10:55:34 +0000 (10:55 +0000)]
1.0.23.35: CLOS tweaking

 * Muffle undefined function style-warning for slot-accessors. The
   slot-accessor name is an internal detail, and defining functions
   accessing slots before classes with such slots are defined is fine.

 * Don't double-fetch slot-definition-initfunction in
   SHARED-INITIALIZE (SLOT-OBJECT).

 * Don't double-fetch various slot-definition properties in
   COMPUTE-EFFECTIVE-SLOT-DEFINITION-INITARGS, and declare
   type-check-functions as functions.

15 years ago1.0.23.34: A nicer register dump for ldb.
Thiemo Seufer [Thu, 11 Dec 2008 20:38:32 +0000 (20:38 +0000)]
1.0.23.34: A nicer register dump for ldb.

15 years ago1.0.23.33: Stack-allocatable vectors for MIPS.
Thiemo Seufer [Thu, 11 Dec 2008 20:28:13 +0000 (20:28 +0000)]
1.0.23.33: Stack-allocatable vectors for MIPS.

15 years ago1.0.23.32: Flush icache for MIPS alien callback stub.
Thiemo Seufer [Thu, 11 Dec 2008 20:26:00 +0000 (20:26 +0000)]
1.0.23.32: Flush icache for MIPS alien callback stub.

15 years ago1.0.23.31: NFP is not a descriptor-reg.
Thiemo Seufer [Wed, 10 Dec 2008 22:11:51 +0000 (22:11 +0000)]
1.0.23.31: NFP is not a descriptor-reg.

15 years ago1.0.23.30: Don't export storage class names.
Thiemo Seufer [Wed, 10 Dec 2008 22:09:59 +0000 (22:09 +0000)]
1.0.23.30: Don't export storage class names.

15 years ago1.0.23.29: Specify the required double-stack alignment for MIPS.
Thiemo Seufer [Wed, 10 Dec 2008 22:07:06 +0000 (22:07 +0000)]
1.0.23.29: Specify the required double-stack alignment for MIPS.

15 years ago1.0.23.28: defconstant -> def!constant in the MIPS backend.
Thiemo Seufer [Wed, 10 Dec 2008 22:04:13 +0000 (22:04 +0000)]
1.0.23.28: defconstant -> def!constant in the MIPS backend.

15 years ago1.0.23.27: Less magic constants in the MIPS backend.
Thiemo Seufer [Wed, 10 Dec 2008 22:01:07 +0000 (22:01 +0000)]
1.0.23.27: Less magic constants in the MIPS backend.

15 years ago1.0.23.26: AMOPly correct defaulting of direct superclasses
Nikodemus Siivola [Fri, 5 Dec 2008 13:47:37 +0000 (13:47 +0000)]
1.0.23.26: AMOPly correct defaulting of direct superclasses

  "The class standard-object is the default direct superclass of the
  class standard-class. When an instance of the class standard-class
  is created, and no direct superclasses are explicitly specified, it
  defaults to the class standard-object."

  "The same is true for funcallable-standard-class and
  funcallable-standard-object."

  * Add :DEFAULT-INITARGS to that effect.