- 11 Nov, 2012 1 commit
-
-
- 10 Nov, 2012 2 commits
-
-
Go Devils!
-
DEFSYSTEM-DEPENDS-ON loads its requirements before the system definition is evaluated which is not needed in this case. The abcl-test-lisp ASDF definition is contained in the toplevel "abcl.asdf" file, which I usually have symlinked into ~/.asdf-install-dir/systems/ for convenience.
-
- 09 Nov, 2012 7 commits
-
-
Added non-working test to putative JSS test suite.
-
We allow DEVICE lists to contain a string value as constructed by MAKE-PATHNAME, but the result can never actually be resolvable by TRUENAME. Instead of trying to figure out the proper use of Java labels, just use the private static Pathname.doTruenameExit() as the common point for all exits from the TRUENAME implementation.
-
-
An ASDF:MVN component can now optionally specify a CLASSNAME that if able to be found in the current jvm process, inhbits further loading from the network. An ASDF:MVN component may also optionally specify an ALTERNATE-URI that will be added to the current jvm classpath if Maven cannot be invoked. Most jvm implementations won't access such jar archives from the network by default, so this is an intermediate step before actually writing the code to download the jar to the local filesystem to then be added. Refresh documentation in README.markdown.
-
-
-
Further work needed as something more meaningful should be returned other than "The value NIL is not of type REAL" when abcl-asdf can't parse the Maven version string. Initial patch by Christoph.
-
- 07 Nov, 2012 1 commit
-
-
Thanks for Christoph.
-
- 02 Nov, 2012 2 commits
-
-
-
Need to get ABCL differences back upstream.
-
- 31 Oct, 2012 1 commit
-
-
- Followup to #14216
-
- 30 Oct, 2012 1 commit
-
-
This fixes the immediate problem, but leave the ticket open as far as I can tell the following code fails in ABCL, but succeeds in the equivalent Java code indicating we need to fix Java.isApplicableMethod(): (jstatic "asList" "java.util.Arrays" (java:jnew-array (jclass "int") 1))
-
- 29 Oct, 2012 3 commits
-
-
-
-
http://beta.quicklisp.org/quicklisp.lisp
THe JVM seems confused with the AMS CloudFront certificate wildcards as to whether 'https://beta.quicklisp.org' is valid or not, so switch to using "https". N.b. ASDF loading of QUICKLISP-ABCL currently doesn't trigger the load-op, probably because we have no declared artifacts, so this contrib cannot be considered as "working" yet. But if the ASDF:PERFORM *were* to get called, things should work.
-
- 27 Oct, 2012 2 commits
-
-
Enables a (to be released) patched CFFI to actually use callbacks.
-
-
- 25 Oct, 2012 1 commit
-
-
The use of "-" as a separator for fields in imitation of GNU configure would probably be better replaced with returning proper lists.
-
- 24 Oct, 2012 2 commits
-
-
- 23 Oct, 2012 1 commit
-
-
Added docstrings for socket API as exported by EXTENSIONS.
-
- 21 Oct, 2012 5 commits
-
-
Remove debugging code from previous work.
-
-
-
due to not being unread into underlying stream.
-
-
- 20 Oct, 2012 2 commits
-
-
-
- we used cl:function as slot name, which is not ANSI-compliant.
-
- 19 Oct, 2012 1 commit
-
-
Fix logic for working around a missing Maven. ABCL-ASDF:RESOLVE had a broken conditional.
-
- 17 Oct, 2012 2 commits
-
-
ABCL-ASDF now knows how to load a locally installed Quicklisp. To download and install Quicklisp from a clean REPL: CL-USER> (require :abcl)(require :abcl-contrib)(require :quicklisp-abcl)
-
FIXME: Currently need to force load via: (asdf:load-system :abcl-contrib :force t)
-
- 16 Oct, 2012 1 commit
-
-
TODO: develop a "lazy serialization" so that all of the arguments of SHA256 will have the underlying java.nio.Channel objects ready t obe run through the digest function. Enough for an implementation of renaming dynamic callbacks.
-
- 15 Oct, 2012 2 commits
-
-
-
SHA256 only working for PATHNAME objects at the moment.
-
- 14 Oct, 2012 3 commits
-
-
- sha256 was corrupted (truncated line in typecase form), kludged into compiling without trying to guess what the form was supposed to be
-
-
-