Revisions of sbcl
Dominique Leuenberger (dimstar_suse)
accepted
request 441531
from
Ruediger Meier (rudi_m)
(revision 42)
1
Dominique Leuenberger (dimstar_suse)
accepted
request 436422
from
Ruediger Meier (rudi_m)
(revision 41)
1
Dominique Leuenberger (dimstar_suse)
accepted
request 414946
from
Matthias Mailänder (Mailaender)
(revision 40)
1
Stephan Kulow (coolo)
accepted
request 316612
from
Togan Muftuoglu (toganm)
(revision 38)
- Update to version 1.2.13 * Incompatible change: + on success, TRY-SEMAPHORE and WAIT-ON-SEMAPHORE return the new count * Enhancement: + WAIT-ON-SEMAPHORE accepts a decrement parameter + JOIN-THREAD allows distinguishing timeout vs. abort in all situations + On Windows DBG_PRINTEXCEPTION_C is handled and its message is printed. (#1437947) * Bug-fix: + TRUENAME works properly on broken symlinks presented as directories. (#1458164) + Inlined DPB and DEPOSIT-FIELD don't interfere with left-to-right order of argument evaluation. (#1458190) + (SETF (LDB (BYTE 1 2 JUNK) X) 0) is rightly rejected. + DEFSETF lambda lists should not permit argument destructuring. + calls to (SETF SLOT-VALUE) on a missing slot would in certain situations incorrectly return the result of a SLOT-MISSING method instead of always returning the new value. (#1460381) + a DEFMACRO occurring not at toplevel and capturing parts of its lexical environment (thus being a closure) caused expressions involving the macro name to cause corruption in the pretty-printer due to faulty introspection of the lambda list of a closure. + out of line MAP/MAP-INTO check that the results produced by the function are of the matching sequence type. (#1459581) + pretty-printing of '(LET `((,X ,Y)) :B) is handled correctly. (forwarded request 316611 from toganm)
Dominique Leuenberger (dimstar_suse)
accepted
request 313380
from
Factory Maintainer (factory-maintainer)
(revision 37)
Automatic submission by obs-autosubmit
Dominique Leuenberger (dimstar_suse)
accepted
request 305183
from
Togan Muftuoglu (toganm)
(revision 36)
- Update to version 1.2.11 * Enhancement: + SET-PPRINT-DISPATCH will warn when given an expression in which any part is unrecognizable as a legal type-specifier. The dispatch table will be altered, but the new entry is disabled. Subsequent type-defining forms will cause pprint-dispatch tables to re-examine whether any disabled entries should be enabled. (lp#1429520) + Loading code containing calls to a deprecated function will, under most circumstances, signal warnings similar to compiling such code. The usual caveat holds about not detecting calls through a computed name, as in (funcall (intern "DEPRECATED-FUN" "SB-EXT")). + (SB-EXT:COMPILE-FILE-LINE) is a new macro that expands to a constant (VALUES integer integer) indicating the source line/column from which it was read, intended for logging Lisp runtime errors in a style similar to that afforded by the C preprocessor __LINE__ macro. Similarly (SB-EXT:COMPILE-FILE-POSITION) returns a position in characters. + improved source locations for VOPs, alien types and declarations. * Bug fix: + Functions in :FINAL deprecation have the correct docstring. No visible change, as no such functions presently exist. (lp#1439151) + (SETF (FDEFINITION this) (FDEFINITION OTHER)) signals an error if OTHER names either a macro or special-operator. (lp#1439921) - Rebase all patches. (forwarded request 305182 from toganm)
Dominique Leuenberger (dimstar_suse)
accepted
request 294078
from
Togan Muftuoglu (toganm)
(revision 35)
- Update to version 1.2.10 * One thing that is not mentioned in the release notes, but which has been exercising developers recently, is a warning for the near future: some interfaces which are currently in "early" deprecation status (and so are emitting style-warnings at compilation-time) will in the next release cycle or two start emitting full warnings, either at compilation time or at load time. Particular interfaces signalling warnings causing problems to some users include: SB-EXT:QUIT, and the SB-THREAD:SPINLOCK API. If anyone is using those, now is the time to update to SB-EXT:EXIT and SB-THREAD:MUTEX respectively. (See also the "Deprecated Interfaces" chapter in the manual). * minor incompatible change: + all SOCKINT::WIN32-* functions have been deprecated with EARLY deprecation state + performing introspection via the system-internal SB-INT:INFO function could expose that :TYPE :TRANSLATOR is not necessarily a function, as it always was before. (Affects swank-fancy-inspector) * enhancement: + The value of SXHASH on bit-vectors of length equal to the word size now depends on the vector's contents instead of being constant; its value on bit-vectors of length divisible by the word size now depends also on the contents of the vector's last word. * bug-fix: + sb-bsd-sockets on win32 uses proper C function declarations. (lp#1426667) + A new dead code elimination phase removes dead code loops that confuse stack analysis. (lp#1255782, lp#308914) (forwarded request 294077 from toganm)
Dominique Leuenberger (dimstar_suse)
accepted
request 288296
from
Togan Muftuoglu (toganm)
(revision 34)
- Update to version 1.2.9 * Bug Fix: + minor incompatible change and bug fix: unboxed numeric constants on x86oids are arranged in memory differently, and the disassembler does not show them separately in DISASSEMBLE, but does if DISASSEMBLE-CODE-COMPONENT is used. (lp#1421987) + DEFCLASS handles cyclic {super,meta}class relations better (lp#1418883) + compiler no longer signals an error when compiling certain function calls. (lp#1416704, lp#404441, lp#1417822, lp#1234919) + compiler doesn't stumble on a LOGIOR transform. (lp#1389433) + more robust debugger and backtraces. (lp#1413850, lp#1099500, lp#1412881, lp#945261, lp#1419205, lp#1409402) + files larger than 4GB can now be compiled. + x86 truncated results from 64-bit foreign functions to 32 bits. + file-position didn't work on large files on win32. (lp#1271545) + callbacks from foreign threads can work without enabling sb-safepoint. + sb-introspect:function-lambda-list works properly on interpeted macros. (lp#1387404) + ADJUST-ARRAY properly handles non-adjustable arrays. (lp#886418) + compiler no longer fails to dump a multidimensional array constant involving a circular reference to itself + conditional and nested DX allocation no longer confuse the compiler in STACK analysis. (lp#1044465) + sb-rotate-byte constant folding bug fixed. (lp#1423682) (forwarded request 288295 from toganm)
Dominique Leuenberger (dimstar_suse)
accepted
request 283726
from
Togan Muftuoglu (toganm)
(revision 33)
- The release (source) tarball of sbcl-1.2.8 contained an error in the build script, which in some circumstances (chiefly when /bin/sh is bash) would lead to the sbcl built from that tarball to have a lisp-implementation-version of "-dirty". (forwarded request 283724 from toganm)
Dominique Leuenberger (dimstar_suse)
accepted
request 283620
from
Togan Muftuoglu (toganm)
(revision 32)
- Update to version 1.2.8 * Enhancement: +better error and warning messages. (lp#1314767, lp#736383) + backtrace for invalid argument count produces the exact supplied arguments including the extra ones, on x86, x86-64 and ARM. + STYLE-WARNING is signaled for DEFSTRUCT accessors which are used before the structure is defined; similarly for the predicate. * Optimization + FORMAT NIL with only ~a and string arguments is transformed into CONCATENATE. + POSITION and FIND when inlined into code that is compiled with qualities of safety < 2 and speed > space will no longer signal an error on circular lists, but will potentially loop forever if given no :END constraint. As was always the case, calls that are not inlined are safe regardless of lexical policy. * Bug Fix: + CLOS methods compiled with (OPTIMIZE (DEBUG 0))no longer cause debugger failure when printing a backtrace + more resilience to deleted code. (lp#1308328, lp#1390544) + the CLHS example of MAKE-LOAD-FORM involving TREE-WITH-PARENT did not work, and now it does. (forwarded request 283619 from toganm)
Dominique Leuenberger (dimstar_suse)
accepted
request 280515
from
Togan Muftuoglu (toganm)
(revision 31)
- Update to version 1.2.7 * Optimization: + returning constant values refers to preboxed constants more reliably. (lp#1398785) * Enhancement: + STYLE-WARNING is produced if a compiler-macro is defined for a function after at least one ordinary (not inlined) call to that function was compiled, indicating a likely compilation order problem. Likewise a warning ensues if a call is compiled to a function that is subsequently proclaimed INLINE. + always lose() when something goes wrong while saving a core (instead of just printing an error on stderr in some situations). + frames in the debugger are now restartable by default. * Bug Fix: + restore error handling on Windows x86. + MAKE-SEQUENCE detects type errors in its :INITIAL-ELEMENT at compile-time when possible. (lp#330299) + parsing of malformed type specifiers no longer results in a memory-fault-error. + LOOP properly destructures nested lists in WITH. + MACROEXPANDing the redefinition of an alien structure type no longer signals the wrong error. + PROGV doesn't get confused by forced DEBUG 3 (lp#1405456). (forwarded request 280514 from toganm)
Dominique Leuenberger (dimstar_suse)
accepted
request 263737
from
Togan Muftuoglu (toganm)
(revision 30)
- Update to version 1.2.6 * enhancement: + SERVE-EVENTS uses the poll() system call in lieu of select() if the OS has the former. Previously poll() was used only if waiting on exactly one file descriptor. + efficiency of access to untagged structure slots is improved on x86-64, and the order of slots in memory is exactly as specified by defstruct, simplifying use of structures as arguments to foreign calls. * bug fix: + SB-DEBUG:ARG now works in all TRACE options which evaluate forms. (lp#1357826) + GC memory corruption during internal memory handling. + duplicate effective-slot-definition objects as compared by EQ on name could be present in CLASS-SLOTS of a class whose metaclass was structure-class or condition-class. (lp#1049423) + HANDLER-BIND with empty bindings works again; regression in 1.2.5. (lp#1388707) + ATOMIC-INCF works on structure slots in interpreted code. (lp#1381867) + MAKE-ARRAY properly handles character types like (eql #\a) and (member #\a #\c). (lp#1392068) + bug fix: READ sometimes accidentally preserved a whitespace character after a token when it should not have. (lp#327790) (forwarded request 263736 from toganm)
Dominique Leuenberger (dimstar_suse)
accepted
request 261437
from
Factory Maintainer (factory-maintainer)
(revision 29)
Automatic submission by obs-autosubmit
Stephan Kulow (coolo)
accepted
request 259750
from
Togan Muftuoglu (toganm)
(revision 28)
- Update to version 1.2.5 * Enhancement + sb-bsd-sockets now has basic support for IPv6 + An sb-unicode package has been added, containing many functions related to handling Unicode text + The reader now normalizes symbols to Normalization Form KC (NFKC). This behavior can be disabled with SB-EXT:READTABLE-NORMALIZATION + enhancement: a style-warning is signaled if OPTIMIZE declarations multiply specify a quality with differing values. (lp#310267) * Bug fix: + conservatively pointed to pages wipe out unused dwords so that they cannot act as false roots in turn. + the walker's handling of lexical variable and symbol-macro bindings is improved (lp#375326, lp#1368305) + HANDLER-{BIND,CASE} no longer drop into ldb when a clause contains an undefined condition type; regression in 1.1.19( lp#1378939) + in interpreted code, inequality predicates did not type-check arguments that weren't examined, and a 1-argument use of MIN or MAX accepted a complex number. (lp#1373702) + APROPOS and APROPOS-LIST handle inherited symbols correctly. (lp#1364413, thanks to Zach Beane) (forwarded request 259748 from toganm)
Stephan Kulow (coolo)
accepted
request 247008
from
Togan Muftuoglu (toganm)
(revision 27)
- Update to version 1.2.3 * Enhancement: + DOCUMENTATION works on instances of FUNCALLABLE-STANDARD-CLASS + ASSERT reports arguments of calls to lexical functions in additional to global functions. + ASDF updated to 3.1.3 + UIOP can be loaded without ASDF. * Bug Fix: + Some methods on CHANGE-CLASS did not finalize the new class before accessing its class-precedence list. + CHANGE-CLASS no longer fails on unbound slots in the source object. + CHANGE-CLASS signals an appropriate error when the destination class is a FORWARD-REFERENCED-CLASS. + CHANGE-CLASS no longer signals bogus TYPE-ERRORs for slots for which initargs have been supplied. + saving cores on ARM was broken since the introduction of GENCGC. (lp#1349795) + macroexpanding a backquoted expression involving ",." outside of the compiler failed. (lp#1354623) - Backported from git for arm builds 0002-Fix-array-bound-checking-with-negative-indexes-on-AR.patch (forwarded request 247007 from toganm)
Adrian Schröter (adrianSuSE)
committed
(revision 26)
Split 13.2 from Factory
Ludwig Nussel (lnussel)
accepted
request 244289
from
Togan Muftuoglu (toganm)
(revision 25)
- disable-localport-bsd-sockets-test.patch * disable localport bsd sockets tests broken in kvm builds for openSUSE releases > 13.1 (forwarded request 244288 from toganm)
Stephan Kulow (coolo)
accepted
request 239067
from
Togan Muftuoglu (toganm)
(revision 24)
- Update to version 1.2.1 * Enhancement + GENCGC is enabled on ARM. + better error reporting for invalid calls to local functions. + support for GNU/kFreeBSD distributions. (lp#1079954, thanks to Christoph Egger) + experimental support for threads on NetBSD/x86-64. (thanks to Robert Swindells) + support for DragonFly BSD. (lp#1292845, thanks to Vasily Postnicov) * Bug Fix: + TYPE-OF must not return AND/OR/NOT expressions. (lp#1317308) + accessing NIL arrays stopped producing errors. (lp#1311421) + DISASSEMBLE no longer prints unwanted stuff. (lp#1251861) + compiling SVREF on unknown types no longer produces scary errors. (lp#1258716) + assorted LOOP fixes and enhancements. (lp#645534, lp#1322923, lp#700538, lp#613876, lp#695286, lp#798388) + silence compiler notes from the SEARCH transform. (lp#1071310) + array type intersection handles T and unknown element types correctly. (lp#1258716) + fix a corner case in array type unparsing that would result in misleading translations from our internal type representation. + array-rank now sees through union and intersection types. (lp#1310574) + when DECLARE expressions are in the wrong place, be careful not to attempt to evaluate the subforms. (lp#573747; thanks to Roman Marynchak) (forwarded request 239066 from toganm)
Displaying revisions 61 - 80 of 102