DEFSTRUCT shouldn't issue warnings for FBOUNDP conflicts at macroexpansion time. (at compile time or load time, OK, but not macroexpansion time) Make DESCRIBE on a gf report its method combination. manual fix from Brian Mastenbrook sbcl-devel 16 Jun 2004
Fix bug in INTERRUPT-THREAD ... pin the function so it can't be moved by GC; ... not /entirely/ clear why it's necessary, but...
Fix ridiculous slowness (and verbosity) of sb-posix STAT/TERMIOS ... construction is still a bit baroque.
Oops. Restore build-from-scratch-ability. ... only style-warn on the target, to spare our blushes and also because STYLE-WARN doesn't work that early. ... (possibly not the optimal fix, but it was certainly the fix that required least thought)
Fix bug 276. Woo yay. Now we can be evil in DEFMETHODs again. ... also log a couple more HaibleMOPBugs
There is no excuse for &OPTIONAL and &KEY arguments in the same lambda list. Really not. (Silly ANSI, silly CLIM) ... STYLE-WARN on function lambda lists with this bogosity; ... tests.
OK, OK, you (the hordes of complaining users) win. ... no more optimization notes in default compilation policy.
Better EXPT behaviour ... or at least different EXPT behaviour. Arrange to (attempt to) return a floating point 1, rather than a NaN, when the base is NaN and the exponent is zero.
Better %UNARY-FTRUNCATE behaviour ... 158 failures (out of 21597, version 1.4) ... woohoo. "Two nines" compliance!
Implement a slightly-broken %UNARY-FTRUNCATE ... slightly broken because it doesn't distinguish between positive and negative zeros ... however, it's better than before: x86/Linux is now down to 232 failures on ieeefp-tests 1.4 ... will fix the brokenness shortly
Delete some unused vops from x86 float backend in preparation for some more refactoring. ... still 392 test failures
Fix errant #+sb-doc that was causing CLEAR-INFO to have no body when docstrings are disrequested.
Less noise from the compiler ... Remove :IMPORTANT T from %COERCE-CALLABLE-TO-FUN transform, so that it complains about uncertain type only with (OPTIMIZE SPEED). ... Test.
Fix countless bugs in backquote printing. ... descend quoted list structure, necessary in nested backquotes; ... fix the fix to Brian Downing's bug: MAPCAR is not sufficiently like MAPCAN. ... add a couple of tests, but frankly we need some more. If someone out there has a test suite for backquote behaviour, that would be rather nice.
Storing a floating point control word on the stack, then calling out to C, is not guaranteed to behave when the called function is likely to be messing around with the stack (purify() in particular). ... reload the floating point control word before setting the precision. This differs slightly in semantics from before: changes caused by the C code will be propagated to Lisp
FINALLY! Fixed bugs related to DOUBLE-FLOAT-EPSILON on x86. Die, bug #45, die a horrible death. ... make lisp code run with the fpu set to 53-bit mantissa; ... add code in number stack allocation to set the fpu to 64-bit precision when calling out to C, conditional on new optimization quality SB-C::FLOAT-ACCURACY (unexported, undocumented, etc) ... use FLOAT-ACCURACY around syscalls, since they don't involve the FPU. Also add code to allow fpu precision control in the (nominally private) sb-int:set-floating-point-modes. (this checkin fixes not only bugs #45 and #118, but also six of PFD's EPSILONS tests and 29 ieeefp-tests related to +, -, *, / and sqrt)
Fix backquote pretty-printing bug (reported by Brian Downing sbcl-devel 2004-06-06) ... don't destroy list structure in the process of printing!
Make ED customizeable, similar to REQUIRE ... ED functions should attempt to implement the standard for ED where practical; ... point to documentation when ED or REQUIRE fail.
Fixed bugs in ATANH ... (atanh #c(1 <anything>)) now computes the right answer; ... (atanh <large>) now does the right thing too.
Another, really-a-bug variant of #328. Reported by Tony Martinez.
Installation related fixes ... Include SUPPORT in binary distros, since install.sh now installs it. ... Align mis-aligned column in INSTALL: OpenBSD works on x86, not PPC.
Fix typo: customize-target-features.lisp-expr -> *.lisp
sb-grovel & sb-posix symbol export changes ... sb-grovel no longer quietly exports every constant ... instead, I restored the old default of not exporting the symbol (and, additionally:) unless asked to ... sb-posix now exports all groveled-for constants and structure/class accessors ... (implicitly:) sb-bsd-sockets now does no more export its constants. This corresponds to the behavior before 0.8.11.65.
Squish bug #330 ... PARSE-NAMESTRING now accepts any valid pathname designator as its defaults argument. ... Poke at it with a regression stick.
Log a bug (related to fixed bug #327)
Sneaking in under the wire before the door is bolted ... it's a pure bugfix anyway: make the CONTROL-ERROR in FIND-RESTART-OR-CONTROL-ERROR be printable
More BUGS updates: close 2, add 3. ... Close #11 -- fixed nowadays. ... Close #241 -- notabug. ... Add profiling suboptimality from #241 as #328. ... Add sequential defclass bug as #329. ... Yay, a pathname bug, #330
One more piece of unsupported speculative functionality: merge CSR's deboostrap idea for find-package to better support tools which want to reason about source code. ... I believe this should be enough to allow SLIME & friends to present a seamless editing environment for sbcl source code. Maybe.
Export socket conditions and accessors; improve EAGAIN support ... Export all socket errors and the parent condition SOCKET-ERROR from the sb-bsd-sockets package ... Also export socket-* accessors (except socket-stream) ... We now return NIL from socket-connect and socket-receive when connect(2) or recvfrom(2) fails with errno set to EAGAIN. These fixes were suggested by Tony Martinez.