[armedbear-devel] add fasl version to default directory for asdf cache
Otherwise one gets the like of: Error loading /Users/alanr/repos/lsw2/trunk/scripts/system-registry.lisp at line 42 (offset 1356) #<THREAD "interpreter" {7A8516}>: Debugger invoked on condition of type SIMPLE-ERROR Incorrect fasl version: jar:file:/Users/alanr/.cache/common-lisp/armedbear-0.20.0-dev-darwin-unknown/Users/alanr/repos/lsw2/trunk/jss/invoke.abcl!/invoke._ Restarts: -Alan
Here's the proposed change to asdf.lisp Adds fasl version to the implementation string, and java-1.4 - java-1.7 to the architecture features. This gives (asdf::implementation-identifier) -> "armedbear-0.20.0-dev-fasl36-darwin-java-1.5" -Alan Index: /Users/alanr/repos/abcl/src/org/armedbear/lisp/asdf.lisp =================================================================== --- /Users/alanr/repos/abcl/src/org/armedbear/lisp/asdf.lisp (revision 12660) +++ /Users/alanr/repos/abcl/src/org/armedbear/lisp/asdf.lisp (working copy) @@ -2297,7 +2297,8 @@ (defparameter *architecture-features* '((:x86-64 :amd64 :x86_64 :x8664-target) (:x86 :i686 :i586 :pentium3 :i486 :i386 :pc386 :iapx386 :x8632-target :pentium4) - :hppa64 :hppa :ppc64 (:ppc32 :ppc :powerpc) :sparc64 :sparc)) + :hppa64 :hppa :ppc64 (:ppc32 :ppc :powerpc) :sparc64 :sparc + :java-1.4 :java-1.5 :java-1.6 :java-1.7)) (defun lisp-version-string () (let ((s (lisp-implementation-version))) @@ -2329,7 +2330,8 @@ #+lispworks (format nil "~A~@[~A~]" s (when (member :lispworks-64bit *features*) "-64bit")) ;; #+sbcl (format nil "~a-fasl~d" s sb-fasl:+fasl-file-version+) ; fasl-f-v is redundant - #+(or armedbear cormanlisp mcl sbcl scl) s + #+armedbear (format nil "~a-fasl~a" s SYSTEM::*FASL-VERSION*) + #+(or cormanlisp mcl sbcl scl) s #-(or allegro armedbear clisp clozure cmu cormanlisp digitool ecl gcl lispworks mcl sbcl scl) s)) On Sun, May 9, 2010 at 12:13 AM, Alan Ruttenberg <alanruttenberg@gmail.com> wrote:
Otherwise one gets the like of:
Error loading /Users/alanr/repos/lsw2/trunk/scripts/system-registry.lisp at line 42 (offset 1356) #<THREAD "interpreter" {7A8516}>: Debugger invoked on condition of type SIMPLE-ERROR Incorrect fasl version: jar:file:/Users/alanr/.cache/common-lisp/armedbear-0.20.0-dev-darwin-unknown/Users/alanr/repos/lsw2/trunk/jss/invoke.abcl!/invoke._ Restarts:
-Alan
On 5/9/10 9:10 AM, Alan Ruttenberg wrote:
Here's the proposed change to asdf.lisp
Adds fasl version to the implementation string, and java-1.4 - java-1.7 to the architecture features.
This gives (asdf::implementation-identifier) -> "armedbear-0.20.0-dev-fasl36-darwin-java-1.5"
[…] Added as [r12664][1]. [1]: http://trac.common-lisp.net/armedbear/changeset/12664 SLIME does something similar, for which the "unknown" bit for platform always bothered me. Using the the Java version used for compilation makes some amount of sense, even though the FASLs should be identical for all versions of Java, as it's our native compiler doing the bit twiddling. I should probably change SLIME as well. -- "A screaming comes across the sky. It has happened before, but there is nothing to compare to it now."
On Mon, May 10, 2010 at 1:10 AM, Mark Evenson <evenson@panix.com> wrote:
On 5/9/10 9:10 AM, Alan Ruttenberg wrote:
Here's the proposed change to asdf.lisp
Adds fasl version to the implementation string, and java-1.4 - java-1.7 to the architecture features.
This gives (asdf::implementation-identifier) -> "armedbear-0.20.0-dev-fasl36-darwin-java-1.5"
[…]
Added as [r12664][1].
[1]: http://trac.common-lisp.net/armedbear/changeset/12664
SLIME does something similar, for which the "unknown" bit for platform always bothered me. Using the the Java version used for compilation makes some amount of sense, even though the FASLs should be identical for all versions of Java, as it's our native compiler doing the bit twiddling.
The fasls might have references to java classes that are different for java 1.5 than 1.6. -Alan (who has some influence on what bits are twiddled too)
I should probably change SLIME as well.
-- "A screaming comes across the sky. It has happened before, but there is nothing to compare to it now."
_______________________________________________ armedbear-devel mailing list armedbear-devel@common-lisp.net http://common-lisp.net/cgi-bin/mailman/listinfo/armedbear-devel
On 5/10/10 7:20 AM, Alan Ruttenberg wrote: […]
SLIME does something similar, for which the "unknown" bit for platform always bothered me. Using the the Java version used for compilation makes some amount of sense, even though the FASLs should be identical for all versions of Java, as it's our native compiler doing the bit twiddling.
The fasls might have references to java classes that are different for java 1.5 than 1.6.
-Alan (who has some influence on what bits are twiddled too)
Point taken, although I can't quite come up with a real "for instance" in my head, mainly because Java5 and Java6 were supposed to be very close in terms of APIs by design.
I should probably change SLIME as well.
I just did, although it doesn't distinguish by FASL versions. It could, but no other Lisp does at this point so I left it alone. -- "A screaming comes across the sky. It has happened before, but there is nothing to compare to it now."
participants (2)
-
Alan Ruttenberg -
Mark Evenson