April 2015 - armedbear-devel - mailman3.common-lisp.net

[armedbear-devel] Bug report: non-standard metaclass causes aberrant behaviour
by William Hounslow 08 Apr '15

08 Apr '15

Dear maintainers #1 a user-defined allocate-instance method is not called at all #2 slot-missing is called with a slot-name of nil #3 update-dependent is not being applied to its keyword arguments Best regards William "1.2.1" "Java_HotSpot(TM)_Client_VM-Oracle_Corporation-1.7.0_17-b02" "x86-Windows_7-6.1" ;;;#1 CL-USER(2): (defclass quux-class (standard-class) ()) #<STANDARD-CLASS QUUX-CLASS {D2A36B}> CL-USER(3): (defmethod allocate-instance :after ((class quux-class) &rest initargs) (format t "allocated ~S~%" class)) #<STANDARD-METHOD ALLOCATE-INSTANCE :AFTER (QUUX-CLASS) {14A0B92}> CL-USER(4): (defclass quux nil (bar) (:metaclass quux-class)) #<QUUX-CLASS QUUX {50039B}> CL-USER(5): (make-instance *) #<QUUX {B2D899}> CL-USER(6): (inspect (fdefinition 'allocate-instance)) 0 LAYOUT -----------> #<LAYOUT {BEF803}> 1 NAME -------------> ALLOCATE-INSTANCE 2 LAMBDA-LIST ------> (CLASS &REST ...) 3 REQUIRED-ARGS ----> (CLASS) 4 OPTIONAL-ARGS ----> NIL 5 INITIAL-METHODS --> NIL 6 METHODS ----------> (#<STANDARD-METHOD ALLOCATE-INSTANCE :AFTER (QUUX-CLASS) {14A0B92}> #<STANDARD-METHOD ALLOCATE-INSTANCE :BEFORE (CLASS) {18C9692}> ...) 7 METHOD-CLASS -----> #<STANDARD-CLASS STANDARD-METHOD {A801B0}> 8 %METHOD-COMBINATION -> #<METHOD-COMBINATION STANDARD {F167BB}> 9 ARGUMENT-PRECEDENCE-ORDER -> (CLASS) 10 DECLARATIONS -----> NIL 11 CLASSES-TO-EMF-TABLE -> #<EQUAL HASH-TABLE 0 entries, 11 buckets {D15EE4}> 12 %DOCUMENTATION ---> NIL ;;;#2 CL-USER(11): (make-instance 'quux) #<QUUX {F2AE7E}> CL-USER(12): (slot-value * 'baz) #<THREAD "interpreter" {1A2FE4D}>: Debugger invoked on condition of type SIMPLE-ERROR The slot NIL is missing from the class #<QUUX-CLASS QUUX {50039B}>. Restarts: 0: TOP-LEVEL Return to top level. ;;;#3 CL-USER(2): (defclass foo-class (standard-class) ()) #<STANDARD-CLASS FOO-CLASS {14A123F}> CL-USER(3): (defclass foo-dep () ()) #<STANDARD-CLASS FOO-DEP {145B486}> CL-USER(4): (defmethod mop:update-dependent ((class foo-class) (dep foo-dep) &key direct-slots &allow-other-keys)) #<STANDARD-METHOD MOP:UPDATE-DEPENDENT (FOO-CLASS FOO-DEP) {155388B}> CL-USER(5): (defclass foo () (bar) (:metaclass foo-class)) #<FOO-CLASS FOO {25D10A}> CL-USER(6): (mop:add-dependent * (make-instance 'foo-dep)) (#<FOO-DEP {F167BB}>) CL-USER(7): (trace mop:update-dependent) (MOP:UPDATE-DEPENDENT) CL-USER(8): (defclass foo () (bar baz) (:metaclass foo-class)) 0: (MOP:UPDATE-DEPENDENT #<FOO-CLASS {25D10A}> #<FOO-DEP {F167BB}> (:DIRECT-SUPERCLASSES NIL :DIRECT-SUPERCLASSES NIL DIRECT-SLOTS ((:NAME BAR) (:NAME BAZ)))) #<THREAD "interpreter" {886140}>: Debugger invoked on condition of type PROGRAM-ERROR Odd number of keyword arguments. Restarts: 0: TOP-LEVEL Return to top level.

2 1

[armedbear-devel] CLOS for ABCL based on interpreted Closette?
by ubikation _ 08 Apr '15

08 Apr '15

I've been trying to read AMOP and looking at the Closette code. It seems to me that ABCL uses Closette for CLOS... but the original Xerox code seems to explicitly warn against using it for compiled CLOS. I was just wondering if this is perhaps the reason for the performance issues of CLOS in ABCL. Do other CL implementations use Closette? The text is "Load this source file directly, rather than trying to compile it." from here: http://homepage.stat.uiowa.edu/~luke/xls/xlispstat/other/closette/closette.… Thanks!

1 0

[armedbear-devel] Compiler loses lexical macro in (eval-when (:compile-toplevel ...))
by Douglas Katzman 08 Apr '15

08 Apr '15

Hi. This code fragment works with (LOAD "/tmp/foo.lisp") but not (COMPILE-FILE "/tmp/foo.lisp") (eval-when (:compile-toplevel :load-toplevel :execute) (defvar *count* 0) (macrolet ((something (sym) `(setf (get ,sym 'uniqueid) (incf *count*)))) (defun register-it (x) (something x)))) (eval-when (:compile-toplevel :execute) (format t "~&OK so far~%") (register-it :foo)) (eval-when (:compile-toplevel :execute) (format t "~&Got ~D~%" (symbol-plist :foo))) Armed Bear Common Lisp 1.3.0 Java 1.7.0_51 Oracle Corporation Java HotSpot(TM) 64-Bit Server VM Low-level initialization completed in 0.258 seconds. Startup completed in 1.231 seconds. Type ":help" for a list of available commands. CL-USER(1): (compile-file "/tmp/foo.lisp") ; Compiling /private/tmp/foo.lisp ... ; (DEFVAR *COUNT* ...) ; (DEFUN REGISTER-IT ...) OK so far #<THREAD "interpreter" {3A284BEE}>: Debugger invoked on condition of type UNDEFINED-FUNCTION The function SOMETHING is undefined. Restarts: ...

2 1

[armedbear-devel] Problem in CLOS MOP in abcl 1.2.0
by Pascal Costanza 08 Apr '15

08 Apr '15

Hi, There is a problem in the MOP implementation in abcl 1.2.0: It seems that the parameter list for std-find-method-combination has changed, but the corresponding :initform for :method-combination for standard-generic-function has not been updated to reflect this. I now get an error that std-find-method-combination gets the wrong number of arguments in my test suites. I don't have a good idea how to work around this problem in Closer to MOP, so I believe I cannot officially declare support for abcl 1.2.0. Another recommendation: The *features* list currently doesn't provide a way to detect the version of abcl. Most Common Lisp implementation have one convention or the other to encode this information. If I had a workaround for the above problem, detecting the version via *features* would allow me to provide a version of Closer to MOP that works with several versions of abcl. (This may, however, not be as crucial as for the commercial implementations.) Best, Pascal -- Pascal Costanza

2 1

Re: [armedbear-devel] Site is down
by Victor 08 Apr '15

08 Apr '15

On Wed, 09 Oct 2013 16:19:45 +0300, Victor <bobbie(a)ua.fm> wrote: > On Wed, 09 Oct 2013 16:11:34 +0300, Erik Huelsmann <ehuels(a)gmail.com> wrote: > >> Thanks for the heads-up. The site should be up again. > > Hi Erik, > > the server/host is now up on the internet again but instead of a nice Welcome page it shows a directory listing that reminds me of common-lisp.net before its upgrade. The site is back again. Thanks! > > Thanks, > Victor > > >> >> >> >> On Wed, Oct 9, 2013 at 2:17 PM, Victor <bobbie(a)ua.fm> wrote: >> >>> Hi, >>> >>> I noticed that http://armedbear.org/abcl.html points to the >>> Common-Lisp.net project page that redirects to www.abcl.org that is >>> currently down. >>> >>> Is it possible either to restore the site or turn the redirect off? >>> >>> Thanks, >>> Victor >>> >>>

1 0

[armedbear-devel] Trouble trying to get example to work
by ubikation _ 08 Apr '15

08 Apr '15

I've been finding it difficult to try and understand the Java/ABCL interop. Right now I'm trying to get the official ABCL pure lisp calling java example to work to no avail: http://abcl.org/svn/tags/1.2.1/abcl/examples/pure-lisp-to-java/ Even though I have the Main.class in the local directory, and I see it when I 'echo $CLASSPATH' ABCL says "Class not found: Main". I'm using bash, I've run through the ABCL setup twice (although I'm not sure if it's right) but I am absolutely clueless. Is there a way to view/modify the classpath within ABCL?

2 1

[armedbear-devel] ASDF 3.0.3 released
by Robert Goldman 08 Apr '15

08 Apr '15

Dear Lisp hackers, I am pleased to announce the release of ASDF 2.26. Since previous release 2.25, the changes are as follows: * Run-program much improved, with a slight backward incompatibility. See the new documentation about it. * Portability enhanced, with more robust Windows support, plus patches for CCL, CLISP, LispWorks, XCL. * UIOP improved, with a README, more docstrings, a few new functions (println, writeln, stripln, ensure-gethash, vomit-output-stream), and bugfixes (to ensure-function, with-temporary-file, split-string). * Debian installations that don't export XDG_DATA_DIRS will benefit from the fix to split-string, as the bug was basically disabling the default source-registry entry for /usr/share/common-lisp/source// Closes: #723977 * Minor tweaks to the defsystem, mostly to print more information in some situations, less in other situations, but also to better handle systems with secondary names like foo/bar, a fix for user functions in output-translation, and to :version (:read-file-...) * Documentation somewhat updated. This is a very substantial bug fix release, notably on Windows, but has a stable API wrt the previous version. For this reason, CL implementations are strongly encouraged to replace previous bundled versions with 3.0.3. As always, thanks to Faré for what seems like a countless number of fixes. Thanks to Dave Cooper for many hours of testing on Windows, without whom the improvements on that platform would have been impossible. Work will soon begin on the 3.1.0 series, with some additional API features, so please make sure to pull a copy of 3.0.3 instead of the bleeding edge when you are bundling. Best, Robert

1 0

[armedbear-devel] Something is wrong with abcl.org
by Alan Ruttenberg 08 Apr '15

08 Apr '15

Could someone who is familiar with how it should work check?

4 4

[armedbear-devel] Trying to learn abcl-asdf
by ubikation _ 08 Apr '15

08 Apr '15

I am trying to figure out how to use abcl-asdf to automate the classpath issues I've been dealing with manually in slime via (add-to-classpath "/foo/bar/..."). Are there any good examples of projects using it? I have a few jars in target/dependency that I would like to load. Thank you!

2 1

[armedbear-devel] CLtL2 Environment functions
by Marco Antoniotti 08 Apr '15

08 Apr '15

Hello it looks like ABCL does not implement CLtL2 environment functions. (CLtL chapter 8.5). Would you guys consider putting them in (if not in yet?) Thanks -- Marco Antoniotti

2 2