Open Hub
Open Hub

 Settings |  Report Duplicate

1
I Use This!
Inactive

Commits : Listings

Analyzed about 8 hours ago. based on code collected about 8 hours ago.
Sep 05, 2024 — Sep 05, 2025
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
0.pre7.86.flaky7.15: s/walk-home/scavenge-home/ for consistency with old renamings DTC's analysis of the regression test bug is that Python is losing a variable because its home CLAMBDA is in one COMPONENT while the closing-over CLAMBDA is in another COMPONENT, and the one-COMPONENT-at-a-time physical environment analysis isn't up to the challenge. Thus Python deletes the apparently-unused variable and things deteriorate from there. He produced a fix involving special-casing top-level variables so that physenv analysis never deletes them, but neither he nor I was convinced that this covers all the bases. So I'll try another approach, making FIND-INITIAL-DFO treat the home/closure relationship as just as much of a physical dependency as a function call, and so prevent separation of the two CLAMBDAs into different components. Thus... ...s/call-graph/dependency-graph/ to reflect intended new behavior
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.14: trivial changes made while hunting for the regression test bug
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.13: restored *PRINT-PRETTY* default
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.12: (now bootstraps successfully even without :SB-SHOW) made unbound-PRINT-OBJECT hack unconditional
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.11: (now bootstraps successfully, at least when :SB-SHOW to enable the new unbound-PRINT-OBJECT recovery stuff) added #!+SB-SHOW mechanism to catch and recover from unbound-PRINT-OBJECT gotchas in PCL bootstrapping
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.10: (now gets to src/pcl/print-object.lisp in warm init before dying) s/function/fun/ in DEFSTRUCT PPRINT-DISPATCH-ENTRY to match the s/function/fun/ in slot accessor names elsewhere
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.9: (can now do (SB!DI:TOP-FRAME) at cold toplevel prompt without crashing) fixed %DEFAULT-STRUCTURE-UGLY-PRINT call in DEFAULT-STRUCTURE-PRINT
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.8: found that suppressing the pretty printer lets the system limp into warm init (then die around assem-rtns.lisp)
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.7: split REPL out of TOPLEVEL-REPL, partly for style, partly as part of a hack to try in debugging REPL shouldn't need two levels of LOOP.
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.6: (This version makes it all the way through xc, then dies shortly after TOPLEVEL-INIT.) I guess PLACEHOLDER-FENV is more persistent than I realized, so old CMU CL code to modify it is actually important, so SBCL should modify it too.
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.5: (gets through the typep.lisp problem of flaky7.4, now dies in xc of srctran, apparently because the LABELS code is still broken) s/find-lambda-vars/make-lambda-vars/ s/compute-closure/add-lambda-vars-and-let-vars-to-closures/ The old logic in CLOSE-OVER implicitly relied on LEAF-REFS being completely set up before any CLOSE-OVERs were called, else CLOSE-OVER could terminate prematurely in the (MEMBER THING (PHYSENV-CLOSURE)) clause. It looks as though it might even be the cause of the failure in xc of typep.lisp. Since I'm finding it difficult to grok, must less debug the order in which the compiler initializes and mutates things, any order dependency is the enemy, so rewrite it so that it floods more systematically.
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.4: (This version dies with an AVER failure in FIND-IN-PHYSENV again, this time in xc of typep.lisp. Maybe, since I'm now substantially more knowledgeable about PHYSENVs than I was a few weeks ago, this will be easier to debug.) fixed bug which caused target-sxhash problem. PAIRLIS: "The new pairs may appear in the resulting association list in either forward or backward order." Egads. This is why the xc dies in target-sxhash: It's not a problem in &OPTIONAL handling, it's that I used (PAIRLIS ...) in a recent rewrite of LABELS as though it were (MAPCAR #'CONS ...), and the code breaks because PAIRLIS reverses the order on me. So.. ..Rewrite DEF-IR1-TRANSLATOR LABELS more thoroughly, so it loses its dependence on the order of PLACEHOLDER-FENV.
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.3: (This version cross-compiles all the way up to target-sxhash, where it dies with some sort of bug in &OPTIONAL arguments in FLET.) s/physenv-function/physenv-lambda/ PHYSENV-LAMBDA is read-only. PRE-PHYSENV-ANALYZE-TOPLEVEL does COMPUTE-CLOSURE on all the LETs of each CLAMBDA. PREALLOCATE-PHYSENVS-FOR-TOPLEVELISH-LAMBDAS, which is supposed to play the same role, does not. What bozo wrote that?:-( Ahem. So.. ..Since there's never a good reason to call COMPUTE-CLOSURE on CLAMBDA without calling it on its LETs as well, and since I've demonstrated that it's an attractive nuisance, I'll try to make it less attractive, by making COMPUTE-CLOSURE automatically run over all the the LETs, and moving the old COMPUTE-CLOSURE code into an FLET hidden by the new COMPUTE-CLOSURE interface.
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.2: (still dies with the same assertion failure) factored out some idioms.. ..LAMBDA-BLOCK ..LAMBDA-COMPONENT (previously known as CLAMBDA-COMPONENT) s/entry-function/entry-fun/ s/tail-set-functions/tail-set-funs/
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7.1: (still dies with the same assertion failure) reexpressing things in an effort to understand them.. ..renamed DFO-WALK-CALL-GRAPH to DFO-SCAVENGE-CALL-GRAPH ..s/new-function/new-fun/ ..s/reanalyze-function/reanalyze-fun/ ..s/local-call-analyze-until-done/locall-analyze-clambdas-until-done/ ..s/local-call-analyze/locall-analyze-component/ ..s/local-call-analyze-1/locall-analyze-fun-1/
William Harold Newman
More... almost 24 years ago
0.pre7.86.flaky7: (This version dies early in cross-compilation with an assertion failure, perhaps because PRE-PHYSENV-ANALYZE-TOPLEVEL isn't being called on the right stuff.) tried to straighten out function names and debug names, splitting LEAF-NAME into LEAF-SOURCE-NAME and LEAF-DEBUG-NAME and making both SOURCE-NAME and DEBUG-NAME read-only.. ..IR1-CONVERT-LAMBDA gets both :SOURCE-NAME and :DEBUG-NAME keyword arguments, and then IR1-CONVERT-LAMBDA-BODY and IR1-CONVERT-HAIRY-LAMBDA and IR1-CONVERT-INLINE-LAMBDA do too ..defined DEBUG-NAMIFY to support this ..deleted no-longer-used COMPILE-FIX-FUN-NAME (and made mental note that it's probably the reason that old COMPILE got function debug name right even though %COMPILE doesn't) removed no-longer-used PRIMITIVE-TRANSLATOR stuff noticed that LAMBDA-VARS is read-only
William Harold Newman
More... almost 24 years ago
0.pre7.86: s/top-level/toplevel/, to conform with ANSI EVAL-WHEN situation names
William Harold Newman
More... almost 24 years ago
0.pre7.85: verified that LAMBDA-NAME isn't just for debugging, but can break the compiler, too:-(
William Harold Newman
More... almost 24 years ago
0.pre7.84: merged AD patch to remove :GLOBAL-VAR :KIND :CONSTANT (sbcl-devel 2001-10-02)
William Harold Newman
More... almost 24 years ago
0.pre7.83: deleted unused FAST-SYMBOL-FUNCTION and OBJECT-NOT-FUNCTION-OR-SYMBOL-ERROR replaced %COERCE-NAME-TO-FUNCTION and RAW-DEFINITION with %COERCE-NAME-TO-FUN s/coerce-callable-to-function/coerce-callable-to-fun/ removed FSET in favor of (SETF SYMBOL-FUNCTION)
William Harold Newman
More... almost 24 years ago
0.pre7.82: s/static-function/static-fun/ another DEFSTRUCT cleanup.. ..rewrote structure constructor form to try to make it more concise and readable
William Harold Newman
More... almost 24 years ago
0.pre7.81: DEFSTRUCT cleanups.. ..got rid of old SLOT-ACCESSOR-FORM in favor of %ACCESSOR-PLACE-FORM, so that now information about the different types of raw slots is more nearly centralized in *RAW-SLOT-DATA-LIST* ..moved target-only stuff from near %TARGET-DEFSTRUCT call in %DEFSTRUCT into %TARGET-DEFSTRUCT miscellaneous cleanups.. ..added (SETQ *PRINT-CIRCLE* T) to the before-proper-printing hacks in make-target-2.sh fixed stupid *PRINT-CIRCLE*-related bug in OUTPUT-OBJECT (introduced in the 0.pre7.76 changes)
William Harold Newman
More... almost 24 years ago
0.pre7.80: fixed raw slot accessor stuff so that it indexes correctly when it's looking at raw data (rescaling the index by the element size, as the old DEFUN SLOT-ACCESSOR-FORM code did but my new code doesn't) redid *RAW-TYPE->RAWREF-FUN-NAME* as *RAW-SLOT-DATA-LIST* to support this
William Harold Newman
More... almost 24 years ago
0.pre7.79: added %COMPILER-DEFSTRUCT-generated inline expansions for type predicates removed predicate DEFUN from macroexpansion of DEFSTRUCT (so that now %DEFSTRUCT-generated closures and %COMPILER-DEFSTRUCT-generated inline expansions do everything) moved definitions of SPECIFIER-TYPE stuff later in early-type.lisp, after DEFSTRUCT VALUES-TYPE, so that the VALUES-TYPE-P call there can be inlined
William Harold Newman
More... almost 24 years ago
0.pre7.78: miscellaneous DEFSTRUCT-related cleanups.. ..There are no longer DEFSTRUCTs in defstruct.lisp itself, so DEFSTRUCT-parsing stuff no longer needs to be wrapped in EVAL-WHEN. ..moved DEFUN SLOT-ACCESSOR-FUNS to target-defstruct.lisp ..removed lotso /SHOW stuff from defstruct.lisp ..removed REMOVEMEs
William Harold Newman
More... almost 24 years ago
0.pre7.77: removed slot accessor DEFUNs from macroexpansion of DEFSTRUCT (so that now %DEFSTRUCT-generated closures and %COMPILER-DEFSTRUCT-generated inline expanders do everything)
William Harold Newman
More... almost 24 years ago
0.pre7.76: As long as I'm testing whether the old code works as an xc host, I might as well make some pending cleanups.. ..made CERROR use COERCE-CONDITION directly instead of trying to screen it from the already-a-CONDITION case ..renamed stems-and-flags.lisp-expr to build-order.lisp-expr (taking a hint from the explanatory comment at head of file:-) ..factored should-we-CHECK-IT logic out of OUTPUT-OBJECT into COMPOUND-OBJECT-P, and shared it with DUMP-OBJECT ..renamed CIRCULAR-LIST-P to CYCLIC-LIST-P ..factored out UNIQUELY-IDENTIFIED-BY-PRINT-P too ..deleted unused INSTANCE-SET-CONDITIONAL and INSTANCE-XADD stuff
William Harold Newman
More... almost 24 years ago
0.pre7.75: merged MNA READ-SEQUENCE and WRITE-SEQUENCE patch ("Re: .. horks" sbcl-devel 2001-10-23) renamed LISP-STREAM to ANSI-STREAM renamed lisp-stream.lisp to ansi-stream.lisp renamed ANSI-STREAM-related IN-BUFFER-FOO stuff to ANSI-STREAM-IN-BUFFER-FOO renamed +IN-BUFFER-EXTRA+ to +ANSI-STREAM-IN-BUFFER-EXTRA+ renamed ANSI-STREAM-IN-BUFFER-TYPE to ANSI-STREAM-IN-BUFFER (If you have half a dozen namespaces, why not use them?:-|)
William Harold Newman
More... almost 24 years ago
0.pre7.74: merged flaky6 changes back into main branch
William Harold Newman
More... almost 24 years ago
0.pre7.73: added more tests for DEFSTRUCT Don't proclaim/declaim instance types in DEFSTRUCT :TYPE LIST or DEFSTRUCT :TYPE VECTOR (because they're not known to the type system in those cases).
William Harold Newman
More... almost 24 years ago