- 17 Apr, 2014 1 commit
-
-
Assume ASDF has been configured by using the <file:~/.config/common-lisp/source-registry.conf.d/> DSL.
-
- 12 Apr, 2014 1 commit
-
-
32 out of 286 total tests failed: FUNCALL.F-S-P.1, CALLBACKS.LONG, CALLBACKS.UNSIGNED-LONG, CALLBACKS.LONG-LONG, CALLBACKS.UNSIGNED-LONG-LONG, CALLBACKS.POINTER, CALLBACKS.STRING-NOT-DOCSTRING, CALLBACKS.NIL-FOR-NULL, CALLBACKS.FUNCALL.1, CALLBACKS.FUNCALL.2, CALLBACKS.DOUBLE26.FUNCALL, CALLBACKS.FLOAT26.FUNCALL, FOREIGN-GLOBALS.NAMESPACE.1, FOREIGN-GLOBALS.NAMESPACE.2, FOREIGN-GLOBALS.NAMESPACE.4, MAKE-POINTER.HIGH, FOREIGN-ALLOC.1, FOREIGN-ALLOC.2, FOREIGN-ALLOC.3, FOREIGN-ALLOC.4, FOREIGN-ALLOC.6, FOREIGN-ALLOC.8, FOREIGN-ALLOC.9, NULL-POINTER-P.NON-POINTER.1, NULL-POINTER-P.NON-POINTER.2, NULL-POINTER-P.NON-POINTER.3, STRUCT-VALUES-DEFAULT.TRANSLATION.MEM-REF.1, STRUCT-VALUES-DEFAULT.TRANSLATION.MEM-REF.2, FSBV.2, FSBV.3, FSBV.4, FSBV.7. 32 unexpected failures: FUNCALL.F-S-P.1, CALLBACKS.LONG,
-
- 25 Mar, 2014 2 commits
-
-
Under Windows (at least), one has to specifiy "-Djava.library.path=something" after the -classpath option for it to have any effect.
-
Seems to fix loading of Ironclad, and other Quicklisp failures.
-
- 24 Mar, 2014 4 commits
-
-
-
DIRECTORY now works again on UNC paths. UNC paths may be either specified with either back slash (#\\) or forward slash (#\/) doubled as the first character in a Pathname namestring. The patterns in //<server>/<share>/[directories-and-files] are parsed as <server> is stored as HOST. <share> is stored as DEVICE. [directories-and-files] gets parsed as per the normal rules under Windows. Mixing namestrings with both backslash and slash characters can lead to unpredictable results. It is recommended not to use backslash characters in namestrings if it can be avoided. The pathname printed representation is always normalized to using forward slash delimiters.
-
-
From Olof-Joachim Frahm.
-
- 21 Mar, 2014 3 commits
-
-
- 16 Mar, 2014 1 commit
-
-
- 15 Mar, 2014 3 commits
-
-
- 12 Mar, 2014 1 commit
-
-
abcl.rdf is now valid again.
-
- 03 Mar, 2014 1 commit
-
-
- 02 Mar, 2014 2 commits
-
-
-
For intervals less than or equal to a nanosecond, including an interval of zero, the current thread merely yields execution to other threads. Fixes <http://abcl.org/trac/ticket/345>.
-
- 24 Feb, 2014 1 commit
-
-
Instead of calling ABCL-ASDF:RESOLVE in both the ASDF COMPILE-OP and LOAD-OP, we now cache the result of invocation in COMPILE-OP and add this value in the LOAD-OP phase. Contributed by Cyrus Harmon.
-
- 23 Feb, 2014 1 commit
-
-
- 21 Feb, 2014 1 commit
-
-
Contributed by Cyrus Harmon.
-
- 16 Feb, 2014 2 commits
-
-
-
There is apparently a fair amount of "breakage" of cases that used to load no longer working which seems due to the changes in the semantics for finding the FASL init loader. The following tests are now broken but no longer cause the JVM to crash: JAR-PATHNAME.LOAD.HTTP.1, JAR-PATHNAME.LOAD.HTTP.2, JAR-PATHNAME.LOAD.HTTP.4, JAR-PATHNAME.LOAD.HTTP.6, JAR-PATHNAME.LOAD.HTTP.7, and JAR-PATHNAME.LOAD.HTTP.9. Need to follow this up in subsequent work. Fixed the underlying HttpHead.get() interface used to determine whether to used a cache version. The custom HTTP HEAD code that was working under Java 6 no longer worked on Java 7. Added to HttpHead.get() asynchronous java.lang.Throwable on a socket timeout of 5000 ms.
-
- 12 Feb, 2014 1 commit
-
-
- 06 Feb, 2014 2 commits
-
-
-
Fixes Quicklisp, aka the "DIRECTORY no longer works with :WILD-INFERIORS" problem, q.v. <http://abcl.org/trac/ticket/344>. DIRECTORY under non-Windows now fills nil DEVICE components with :UNSPECIFIC, otherwise forms like (equal (truename "~/.emacs") (first (directory "~/.emacs")) ) fail (c.f. ANSI test DIRECTORY.[67]).
-
- 30 Jan, 2014 5 commits
-
-
-
-
-
WILD.PATHNAMES.1 normalization problems were fixed by relaxing semantics on resolving symlinks in CL:DIRECTORY calls.
-
The default for the :RESOLVE-SYMLINKS keyword to CL:DIRECTORY has changed from T to nil, meaning that naked DIRECTORY calls should never return an error for dangling symlinks. Thanks to Marco Antoniotti, Alan Ruttenberg, and Alessio Stallo for clarifying a reasonable position vis a vis ANSI. Addresses <http://abcl.org/trac/ticket/340>.
-
- 27 Jan, 2014 4 commits
-
-
- 26 Jan, 2014 2 commits
-
-
- 25 Jan, 2014 2 commits
-
-
JDK 7(?) changed the default timeout for sockets to 0, which means infinity, which is probably a bad idea when dealing with I/O on the Inner-tubes. Restore the failing tests in ABCL-TEST-LISP: the test suite bombs, but not as catastrophically.
-
Tests are still explicitly chosen via the nested <arg> to the '//abcl.test.java/java@' element at build.xml:940 ff. The test in org.armedbear.lisp.util.HttpHeadTest currently hangs (!) the executing VM, a high priority item for abcl-1.3.0. The problem lies in a hanging read in the JVM system classes that previously didn't exist when accessing the ZipCache.get(url) API for the second time. Users are advised to call SYSTEM:DISABLE-ZIP-CACHE until this behavior is fixed.
-