- Jan 26, 2002
-
-
pw authored
-
- Jan 23, 2002
-
-
toy authored
There is an incorrect gc_assert in lisp/gc.c in the function from_space_p. The argument is not necessarily a descriptor; it may be an untagged pointer, in which case the assertion is false. This causes rebuild to fail on SPARC when GC assertions are enabled. This observation is due to Daniel Barlow (and is fixed in SBCL).
-
toy authored
- remove obsolete (declare (values ...)) forms from the HPPA backend, that were resulting in incorrect code generation
-
toy authored
Update some constants for HPUX.
-
toy authored
other OLD-xxx nicknames go away too?)
-
- Jan 18, 2002
-
-
pmai authored
gratuitous consing caused by coercion at runtime of the ANSI mandated free format boundaries from ratios to appropriate floats. The coercions now happen at compile-time, and documentation is also improved. o Elided duplicate case clause for ratios in output-ugly-object.
-
pmai authored
core. If the newly introduced :batch-mode argument to save-lisp is supplied and is true, then the *batch-mode* flag is set to true, otherwise it is set to false, before saving the core. This should prevent ugly surprises by newbies using -batch when dumping cores.
-
pmai authored
is really exported via the individual export forms. This silences some spurious warnings during compilation.
-
- Jan 16, 2002
-
-
toy authored
SAP, so no need to create an alien object out of it.) Based on a replacement by Lynn Quam.
-
- Jan 15, 2002
-
-
toy authored
- update the parameter FASL-FILE-IMPLEMENTATIONS to include PPC (I presume this was forgotten when merging Gary's PPC changes)
-
toy authored
instead of an (invisible) hyperlink. o Add a section on Lisp threads for x86. (Currently empty. Hope someone who actually uses threads can fill this in.)
-
toy authored
(invisible) hyperlink.
-
toy authored
-
toy authored
-
toy authored
basically conform to ANSI CL on symbols. o Document the FPU precision control feature for x86. o Briefly document that the default random number generator is the MT-19987 generator.
-
toy authored
it look like the document changes all the time!
-
- Jan 14, 2002
- Jan 13, 2002
- Jan 11, 2002
-
-
toy authored
-
toy authored
- the fix to the PPC backend previously discussed here in <Pine.LNX.4.30.0110261852310.13665-100000@clozure.com>: correct the calculation of which offsets can be used in an ADDI instruction
-
toy authored
Bind *tn-ids* and *id-tns* appropriately for apropos so we're not dropped into the debugger.
-
- Jan 10, 2002
-
-
toy authored
files and supporting pdftex and hevea (HTML converter).
-
- Jan 09, 2002
-
-
toy authored
platforms.
-
- Jan 04, 2002
-
-
pw authored
lookup-host-entry fails to find the host. Reported by John Wiseman.
-
- Dec 13, 2001
-
-
pmai authored
because of a missing bounds check.
-
pmai authored
stepping-facility. Since this currently only works on compiled code, we try to compile the anonymous function containing the form to-be-stepped, and bail-out with an error if that fails, e.g. for non-null enclosing lexical environments.
-
pmai authored
COMPILER-MACROEXPAND, and export them from the EXTENSIONS package, since they are useful debugging aids for compiler-macro writers, even if they could write them on their own, and ANSI CL dropped them.
-
- Dec 12, 2001
- Dec 11, 2001
-
-
pmai authored
ensure that CLX is loaded at runtime, too, now that CLX can be loaded via REQUIRE.
-
pmai authored
as uncovered by the recent change to detect this. Folded the larger constants for MINIMUM-WINDOW-LINES and MINIMUM-WINDOW-COLUMNS into the original constant definition, and elided the superfluous defparameter for FONT-MAP-SIZE. This seems to work for now.
-
pmai authored
loaded via REQUIRE.
-
pmai authored
-
pmai authored
names. This change fixes that. For undefined modules, require now defaults to loading "modules:MODULENAME-library", where MODULENAME is treated as if specified in :CASE :COMMON, so that users can use (require :clx) or (require "CLX") to load "modules:clx-library", etc.
-
pmai authored
core without the compiler (e.g. runtime cores). Neither let nor dolist are allowed. The nreverse is there to keep the order of the remaining features unchanged, since set-difference reverses the list in the process.
-
- Dec 10, 2001