- Sep 19, 2002
- Sep 17, 2002
-
-
pmai authored
Hemlock wasn't present, caused the compiler to tail-call eliminate the function call in signal handlers defined via define-signal-handler. Besides causing problems for the debugger, which now sees a foreign function call land frame as the top of stack, this is also confusing to the end user, hence we turn up the debug optimization quality to prevent tail-call-elimination here.
-
- Sep 05, 2002
-
-
toy authored
REAL in the two arg case.
-
- Aug 28, 2002
-
-
pmai authored
o Conditionalized some code that is only used in linkage-table builds on :linkage-table and LINKAGE_TABLE respectively. o Modernized undefined_ff_tramp based on recent changes, and renamed it to undefined_foreign_symbol_trap, because it really isn't a trampoline. o Conditionalized a stray puts(...,dlerror()) in Linux-os.c on DEBUG o Installed an error reporter for UNDEFINED_FOREIGN_SYMBOL_ERROR, which for now signals a simple-program-error. We might want to introduced a new condition UNDEFINED-FOREIGN-SYMBOL or something, which could also be used by the old foreign linkage code.
-
moore authored
Fix foregin object braindamage noted by Pierre Mai.
-
- Aug 27, 2002
-
-
moore authored
On x86 FreeBSD and Linux, change the way foreign symbol addresses are resolved. They now go through a table -- effectively a new space in the core file. Function references are resolved lazily, data references are resolved on startup and when a .so is loaded. The end result is that cores can be dumped that contain references to symbols in shared libraries. Also, the dependence of the core on addresses in the Lisp runtime is broken. The linkage table feature is controlled by :linkage-table and LINKAGE_TABLE in C runtime. Several foreign symbols are now Lisp static symbols, so a cross compile is required whether or not the new stuff is used. I've checked in boot4-cross-foreign-linkage.lisp that builds the compiler for linkage table; do whatever you usually do for the non-linkage table case:) Seriously, lets start a discussion on standardizing "cross compilation," not to mention the general build procedure.
-
- Aug 26, 2002
-
-
toy authored
* distributed binaries should no longer have OLD-xx package nicknames (for the VM and the compiler packages). These nicknames are set up by the build and cross-compile scripts, as needed.
-
- Aug 24, 2002
-
-
pmai authored
than 1024 characters, which also reduces consing for pathnames of more typical lenghts.
-
- Aug 23, 2002
-
-
pmai authored
reporting in several places to be more specific and/or ANSI compliant. The patches have been modified to make use of the new common simple-programm-error function where applicable.
-
pmai authored
-
pmai authored
it defaults to simple-program-error. Removed duplicate definition of simple-program-error from package.lisp, and converted various places to use the new simple-program-error function. Also rewrote the function check-disjoint for clarity.
-
- Aug 22, 2002
-
-
pmai authored
ANSI CL. Also removed several other warnings which were mostly bogus with the existance of writers (the "no :initarg/:initform" warning). And finally also removed the "no :reader" warning, since it is really the perogative of the user to do what he wants, and issuing full warnings is non-conforming to boot.
-
- Aug 20, 2002
-
-
toy authored
From Eric Marsden.
-
- Aug 12, 2002
-
-
toy authored
needs work, but should handle most common cases.
-
toy authored
o CMUCL doesn't have a specialized complex type to hold rationals, so don't return 'RATIONAL. o Return 'REAL instead of T for any type real type. o Give an error if it's something that can't be a component of a complex number.
-
toy authored
have a file-name associated with it, but file-streams do. (Noted by Christophe Rhodes.)
-
toy authored
streamlike as appropriate to catch bad args instead of pretending they're gray-streams to be caught there. o For make-broadcast-stream, make-two-way-stream, make-concatenated-stream, and make-echo-stream, check that their args of input or output streams as appropriate.
-
- Aug 08, 2002
-
-
toy authored
matches the length of the sequence type.
-
- Aug 07, 2002
-
-
toy authored
instead of as the otherwise-clause. Print a simple warning about this being bad style.
-
- Aug 06, 2002
-
-
pw authored
system:read-n-bytes reads short. Problem noted by Todd Sabin.
-
- Aug 02, 2002
-
-
toy authored
bug where read-byte from (signed-byte 8) (or less) streams would produce unsigned instead of signed numbers.
-
- Jul 31, 2002
-
-
toy authored
when redefining structures. Based on suggestions from Rob MacLachlan and Mike McDonald.
-
- Jul 30, 2002
- Jul 29, 2002
-
-
toy authored
(Noted by Bob Rogers.)
-
- Jul 25, 2002
- Jul 22, 2002
- Jul 10, 2002
-
-
toy authored
The search-list "modules:" is set to track the search-list "library:" instead of being set whatever path "library:subsystems/" is set to at start up.
-
toy authored
* (pcl:class-precedence-list (find-class 'null)) should have symbol occurring before list, as per CLHS. * minor changes to certain error conditions: eg trying to set DEFAULT-DIRECTORY to a non-existent directory should raise a condition of type FILE-ERROR. (Some of these changes are from Alexey Dejneka, via SBCL). * cleanup of some Hemlock-related symbols that shouldn't be exported when :no-hemlock is defined.
-
- Jul 06, 2002
-
-
toy authored
o We weren't accepting loops of the form (LOOP FOR I BY (INCF X) FROM ...) (See the example in 6.1.2.1.1 of the CLHS). o PRESENT-SYMBOL means internal and external symbols of a package.
-
- Jun 26, 2002
-
-
pmai authored
connect-to-inet-socket. Since CL byte operations are defined to work on numbers in host order, which might or might not be the same as network order, working portably on network ordered numbers requires conditionalization based on the backend byte-order. Hence it is almost always better to work on host ordered numbers directly.
-
- Jun 19, 2002
-
-
toy authored
error. Catch this in BUILD-RATIO and signal a divide-by-zero error. Based on the bug report and patch by Wolfhard Buss.
-
- Jun 10, 2002
-
-
toy authored
The Hyperspec says that the default for end is nil, but: * (write-string "bla" *standard-output* :start 0 :end nil) Type-error in KERNEL::OBJECT-NOT-FIXNUM-ERROR-HANDLER: NIL is not of type FIXNUM Fix this and make the default for :END be NIL. (Hmm, I suppose we don't really have to default to NIL since NIL means the end of the string anyway, but we should at least accept NIL as a valid :END value. I didn't do that.)
-
- May 08, 2002
-
-
toy authored
%UNARY-FTRUNCATE/DOUBLE-FLOAT as maybe-inline so that they can be inlined by the user if desired.
-
- May 06, 2002
-
-
pmai authored
current Linux and FreeBSD/ELF do, although OpenBSD is still non-ELF. Also changed the handling of leading underscores in alien names, by moving the underscore addition to extern-alien-name, where it belongs, and not foreign-symbol-address-aux. This brings the x86 port in line with the other ports, modulo the PVE_stub_ magic. The changes necessitate some bootstrapping code for BSD non-ELF platforms.
-