Permalink
Switch branches/tags
Commits on Nov 14, 2003
  1. 0.8.5.37:

            * Fix PFD bug MISC.165: unify treating of TAIL-CALLS in
              LET-convertion.
    Alexey Dejneka committed Nov 14, 2003
  2. 0.8.5.36:

            * Fix PFD bugs MISC.166, 167, 168: do not perform tail- and
              let-convertion, if block of lambda's BIND node is to be
              deleted.
    Alexey Dejneka committed Nov 14, 2003
  3. 0.8.5.35:

            * Fix PFD bugs MISC.152, 153: be more accurate with deleted
              code in MAYBE-INFER-ITERATION-VAR-TYPE.
    Alexey Dejneka committed Nov 14, 2003
Commits on Nov 13, 2003
  1. 0.8.5.34

    	Add *INVOKE-DEBUGGER-HOOK*, which works basically like the
    	ANSI *DEBUGGER-HOOK* but isn't bound to NIL in calls to
    	BREAK, so can be used to trap _all_ debugger entry, not just
    	some of it.  Useful for anyone who wants to completely remove
    	the debugger, or supplant it with something else.
    
    	Based on a patch from David Lichteblau, but this one only has
    	a single function instead of a list in the hook.
    telent committed Nov 13, 2003
Commits on Nov 11, 2003
  1. 0.8.5.33:

    	bumped .fasl file version number as per Tim Daly Jr.
    		incompatibility report on sbcl-devel 2003-11-09
    William Harold Newman committed Nov 11, 2003
  2. 0.8.5.32:

            * Fix PFD bug MISC.158: X86 (xor x x) instruction does not
              work for X in memory.
    Alexey Dejneka committed Nov 11, 2003
  3. 0.8.5.31:

    	* contrib/sb-aclrepl/toplevel.lisp: Revert from lexically binding
    	certain state variables.
    kevinrosenberg committed Nov 11, 2003
Commits on Nov 10, 2003
  1. 0.8.5.30

    	"Documentation" update for asdf-install
    
    	Add README file for asdf itself
    telent committed Nov 10, 2003
  2. 0.8.5.29:

    	Completely boring housekeeping commit
    	... rename grovel_headers.c to grovel-headers.c, as that underscore
    		was the only non-logical-pathname-compliant character in
    		the whole of the source and objects filenames.
    	... delete all redundant sb!vm: prefixes throughout
    		src/compiler/$arch/*.lisp
    	... nothing interesting at all, in fact.  Builds and passes tests
    		on at least x86 and sparc.
    csrhodes committed Nov 10, 2003
Commits on Nov 9, 2003
  1. 0.8.5.28:

    	Fix some loop badness (as reported by John Klein sbcl-devel
    	2003-11-09)
    	... step variable guessed initializers should be 1, not 0,
    		because the step type explicitly excludes 0.
    	... log some more loop badness in BUGS
    csrhodes committed Nov 9, 2003
Commits on Nov 8, 2003
  1. 0.8.5.27:

    	SB-{POSIX/BSD-SOCKETS} cleanups
    	... make sb-bsd-sockets-tests depend on sb-posix so we can check
    		to see if /dev/log is actually a socket;
    	... test that /, not /root, is not writeable by others;
    	... (taking advantage of our 0.x version number) alter
    		SB-POSIX:S-ISFOO to use the BOOLEAN alien type, rather
    		than MODE-T -- the point being that the macros have no
    		function other than returning a boolean.  Adjust
    		the sb-posix manifesto too, to document this.
    csrhodes committed Nov 8, 2003
  2. 0.8.5.26:

    	Record filesystem information during the build in debug information
    	relative to SYS, not to /home/kevin/sbcl or equivalent.
    	... change to core code: tiny if ugly;
    	... change to warm load: only to refer to files as lpns rather
    		than physical namestrings;
    	... change to vanilla-modules: tiny;
    	... change to asdf modules: well, erm, less tiny.  I'm not pleased
    		by the need to duplicate similar logic in every .asd, but
    		I think it's a similar symptom to the duplication of the
    		TEST-OP logic in many of them: the coupling between asdf
    		and the rest of the system is slightly wrong.
    
    	(At startup, SYS by default has translations that refer back to
    	the build-time values; distributors may wish to alter this for
    	their binaries, for instance by setting translations to
    	distribution-known values when dumping any final image)
    csrhodes committed Nov 8, 2003
  3. 0.8.5.25:

    	merged Miles Egan's general SGML cleanup manual patch (from
    		sbcl-devel today)
    	deleted "!ENTITY mdash" hack (since ME's patch reminded me that
    		at some point I learned that DocBook itself seems to
    		define mdash)
    William Harold Newman committed Nov 8, 2003
Commits on Nov 6, 2003
  1. 0.8.5.24:

    	merged CSR's well-aged patch to make X86 port stop
    		complaining about register class strangeness
    		in array checking VOPs
    	added #!+ conditionalization on the hack to suppress
    		warnings about analogous strangeness on other
    		ports, hoping I didn't screw up anything I can't test
    William Harold Newman committed Nov 6, 2003
  2. 0.8.5.22:

    	gencgc cleanup:
    	... Replaced all (but one) literal 4096 (and 0xfff) with
    	PAGE_BYTES constant. I think the ones I changed were all about
    	page size, but I added a FIXME where I was unsure. There is
    	one bare 4096 left, but that's not a page size, I think.
    	... Removed gencgc.h, because it wasn't actually needed and
    	mostly duplicated stuff from gencgc-internal.h
    patrikn committed Nov 6, 2003
Commits on Nov 4, 2003
  1. 0.8.5.22:

    	archived some arcane knowledge of porting Hell^W^W^W^W^W wisdom
    William Harold Newman committed Nov 4, 2003
Commits on Nov 3, 2003
  1. 0.8.5.21:

            * Merge David Lichteblau "simple-streams bug with non-SAP
            buffers" patch (sbcl-devel 2003-11-02)
    rudi committed Nov 3, 2003
  2. 0.8.5.20:

    	* contrib/sb-aclrepl/toplevel.lisp: Commit patch from David
    	Lictheblau to better support multiple listener processes.
    kevinrosenberg committed Nov 3, 2003
Commits on Nov 2, 2003
  1. 0.8.5.19:

            * Fix PFD bugs MISC.100, 102, 105, 107, 112;
            ... IR1-MERGE-CASTS: do not put merge asserted type
                contradicting the derived one;
            ... IR1-OPTIMIZE-COMBINATION: try terminate block after
                PROPAGATE-FUN-CHANGE;
            * FLUSH-DEAD-CODE: if the block is split under us, restart.
    Alexey Dejneka committed Nov 2, 2003
  2. 0.8.5.18

    	Minor change to sb-bsd-sockets tests: instead of
    	conditionalising the AF_LOCAL client test on (not (or sunos
    	darwin)), just skip it unless /dev/log exists.  This make life
    	simpler if we're e.g. on a platform that usually has /dev/log
    telent committed Nov 2, 2003
Commits on Oct 30, 2003
  1. 0.8.5.17:

    	Possibly the least useful fix I shall ever make while working on
    	SBCL:
    
    	Implement ldso_stub__foo for MIPS/Linux, so that the system can
    	be built to link dynamically, allowing linking of foreign code
    	in
    	... many thanks to Ralf Baechle for education about the MIPS ABI
    	Now foreign.test.sh passes, and we build and pass tests for all
    	13 contribs!  Woohoo!
    
    	Is there a user in the house?
    csrhodes committed Oct 30, 2003
  2. 0.8.5.16:

    	Aren't we glad we have solaris around to keep us honest?
    	... adjust make.sh message to let it print under real sh(1)
    	... adjust sb-posix tests to let them pass (one true error in
    		the tests, and one weird EINVAL from Solaris)
    csrhodes committed Oct 30, 2003
Commits on Oct 29, 2003
  1. 0.8.5.15:

    	More SB-POSIX stuff
    	... merge patch from Zach Beane (sbcl-devel 2003-10-28) to wrap
    		the stat S_ISFOO macros into callable functions
    	... merge patch CSR sbcl-devel to detect presence of various
    		symbols at compile-time for conditional function
    		definition
    	... slight frobs to make these two coexist: "interface" must
    		depend on "alien", and these calls are NEVER-FAILS
    	... tests from Xach via #lisp IRC
    	... as yet unresolved: duplication of UNIX-DSO stuff
    	Summarize contrib successes at the end of make.sh
    	... hope no-one's trisdekaphobic :-)
    csrhodes committed Oct 29, 2003
  2. 0.8.5.14:

    	Be less assertive about LRAs, since perfectly valid instructions
    	can have LRA widetags.  (Brian Downing sbcl-devel 2003-10-29)
    	... minimally-intrusive and minimally-DWIM patch
    	Define and use NATIVE-FILENAME for sb-posix
    	... also adjust the test not to run RUN-PROGRAM, since we
    		have problems with that :-/
    csrhodes committed Oct 29, 2003
  3. 0.8.5.13:

            G5/PPC970 fix:
    	... Replace all uses of the mcrxr instruction with the mtxer
    	instruction, because mcrxr is reallly slooow on the PPC970 and
    	we didn't actually want it for anything but clearing XER.
    patrikn committed Oct 29, 2003
  4. 0.8.5.12

    	SB-POSIX now converts pathnames to Unix filenames using
    	SB-INT:UNIX-NAMESTRING, instead of to CL namestrings (which
    	are not the same thing, especially when wildchard characters
    	are involved) with NAMESTRING
    
    	Bug report and fix both courtesy of Zach Beane.  Thanks.
    telent committed Oct 29, 2003
Commits on Oct 27, 2003
  1. 0.8.5.11:

    	Modifications for which I could make a plausible case on the
    	basis of being 64-bit SAP implementations for HPPA and MIPS
    	... testing?  What's that?  It compiles, ship it!
    csrhodes committed Oct 27, 2003
  2. 0.8.5.10:

    	PPC support for 64-bit SAP references
    csrhodes committed Oct 27, 2003
  3. 0.8.5.9:

    	SPARC support for 64-bit SAPs, needed for sb-posix
    csrhodes committed Oct 27, 2003
Commits on Oct 26, 2003
  1. 0.8.5.8:

            * Fix bugs MISC.101 and MISC.103 found by Paul Dietz:
              FLUSH-DEST: set BLOCK-TYPE-CHECK attribute for USEs' blocks.
    Alexey Dejneka committed Oct 26, 2003
  2. 0.8.5.7:

            * Fix bug MISC.99 found by Paul Dietz: on X86 in
              FAST-ASH-C/UNSIGNED=>UNSIGNED VOP add case (>= shift 32)
              (the code is shared with -MOD32 version, where "result type
              constraint" does not guarantee overflow absence).
    Alexey Dejneka committed Oct 26, 2003
  3. 0.8.5.6:

    	Fix for LOGAND OPTIMIZER and 0s
    	... use UNSIGNED-BYTE*
    csrhodes committed Oct 26, 2003
  4. 0.8.5.5:

            * Fix bug reported by Brian Downing: do not perform
              MV-LET-convertion, if the last optional entry has
              references.
            ... new consistency condition: function in a local
                mv-combination must be of kind MV-LET.
    Alexey Dejneka committed Oct 26, 2003
  5. 0.8.5.5

    	BUILD_ROOT
    telent committed Oct 26, 2003
Commits on Oct 25, 2003
  1. 0.8.5.3:

    	CLISP minor issues:
    	... fix float-related bug: since CLISP doesn't support
    		denormalized single-floats, we need to construct
    		LEAST-POSITIVE-SHORT-FLOAT explicitly
    	... we need to perform a similar trick for BOOLE as we already
    		do for BYTE.  Rewrite the horrible package hackery to be
    		slightly more robust, in mitigation for the perpetuation
    		of this horror.
    csrhodes committed Oct 25, 2003