gmane.editors.j.devel

Closure.java cleanups, round 10

Ville Voutilainen — 2008-10-07T17:25:01

As mentioned, this patch removes the arg-count (arityValue) parameter from bindParametersAndExecute. No regressions. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ armedbear-j-devel mailing list armedbear-j-devel< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/armedbear-j-devel

Java Scripting API integration for ABCL

Alessio Stalla — 2008-10-06T22:00:17

Hello everyone, As suggested by Ville, since I have a quite stable version of the project, I would ask Erik to integrate it with the ABCL repository in a new branch. The package name is org.armedbear.lisp.scripting. It does not modify any of ABCL's files. There are a couple of minor things which could require modifications to ABCL to work, but they're not very important - I'll explain them in detail on the mailing list as soon as possible. Most of the basic API is implemented, plus 3 of the 4 methods of the optional Invocable interface. There's no documentation and very limited unit testing for now. So, Erik, when you are ready drop me a note and I'll send you the zip. Thanks, Alessio ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

GCL ansi tests on common-lisp.net

Ville Voutilainen — 2008-10-06T18:47:16

Cool. Ok, we can generate a list the items of which are on separate lines? Then we get both the ability to diff and the ability to parse with lisp. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Forthcoming Closure.java cleanups..

Ville Voutilainen — 2008-10-06T17:22:44

On Mon, Oct 6, 2008 at 8:05 PM, Ville Voutilainen gmail.com> wrote: Well, that lasted long. :) I noticed that bindParametersAndExecute compares arity with its parameter, arityValue. This value is always minArgs, so the param can be killed completely and we can just compare arity to minArgs directly. This also shows what the code was attempting to do in the first place. The bindParametersAndExecute function is my doing, but the original code was very unclear with regards to this. So our refactorings help here too, the idea of the code becomes clear(er) after refactoring it. I'll do a patch for this after the previous one (round 9) is committed. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Closure.java cleanups, round 9

Ville Voutilainen — 2008-10-06T17:05:00

Combine bindOptionalParameterDefaults and bindKeywordParameterDefaults into a single function. Also, removes a null check that survived the previous patch, and removes a pointless null thread check from processArgs. At the moment I don't have more cleanups in my queue. I'm planning to start debugging the failing push/pop/pushnew test cases. Before that, have the ansi tests in common-lisp.net svn been fixed so that they run out-of-the-box on abcl? I'd also like to modify the test printout so that each failed test is reported on a separate line, which makes diffing the output easier. -VJV- ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ armedbear-j-devel mailing list armedbear-j-devel< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/armedbear-j-devel

Supporting .lsp file extension?

Erik Huelsmann — 2008-10-05T19:16:11

The ANSI test suite uses .lsp file extensions for Lisp files. ABCL currently uses .lisp. How about supporting .lisp ; looking for .lsp if there's no .lisp? Bye, Erik. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Closure.java cleanups, round 8

Ville Voutilainen — 2008-10-05T16:23:43

I removed the possibility of Parameter and Symbol arrays to be null. This allows for removing a bunch of null checks throughout the code. Net win of 31 code lines, and no regressions. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ armedbear-j-devel mailing list armedbear-j-devel< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/armedbear-j-devel

Proposal: refactor FastStringBuffer for Java 5

Philip Hudson — 2008-10-04T21:56:27

I take FastStringBuffer to be /primarily/ an optimization for avoiding the synchronization overhead of java.lang.StringBuffer. I suggest the following for FastStringBuffer. o Separate out and preserve all existing functionality of FSB over and above that of StringBuffer/StringBuilder. o Replace the guts of the FSB implementation with a composed instance of StringBuilder, which was introduced into the Java 5 API for exactly the same reason as above. This will take out all the state and most of the behavior of the existing FSB. o Implement CharSequence and add a constructor that takes a CharSequence. o Implement Appendable. Gains: o Removal of internally designed, tested and maintained behavior. o Replacement with externally maintained and massively tested functional equivalent. o Conform to Java standard for character sequences. I'm not 100% certain at this point that it's safe and correct to replace the parallel existing FSB behavior with the Java 5 library behavior, though I think it's very likely. I notice that we don't seem to have any Java unit tests. So I propose to start by writing JUnit 4 tests covering FSB, before doing any refactoring, to give a measure of confidence that the refactored class will work correctly. This can act as a best-practice example for extending to eventually cover the whole Java codebase. Of course, we can also include the unit tests in the ant build script, as either an optional or a required target. So, is this the kind of thing we want to be focussing on right now, or is there something higher-priority I should do? -- Phil Hudson PGP/GnuPG ID: 0x887DCA63 ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Closure.java cleanups, round 7

Ville Voutilainen — 2008-10-01T22:19:15

This patch adds the said helper. No test regressions. The patch looks a bit funny, so it's not straightforward to count the line delta from the patch. It still deletes more code than it adds. Next patch will bring the merge of bindOptionalParameterDefaults and bindKeywordParameterDefaults. Then I'll try removing unnecessary null assignments from constructor, and then try avoiding array null checks by ensuring that the arrays are always allocated. So, at the moment it looks like the Closure.java saga will be at least 10 patches long (and somehow I don't think that's the end of it). ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ armedbear-j-devel mailing list armedbear-j-devel< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/armedbear-j-devel

Tests

Philip Hudson — 2008-10-01T20:54:52

Should have asked before submitting my first patch: how do you run ABCL's tests? -- Phil Hudson PGP/GnuPG ID: 0x887DCA63 ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

project page XHTML validation

Alessio Stalla — 2008-10-01T17:22:40

I'm taking this. Seems to be few types of validation errors repeated, it should not take much. Alessio -------------------------------------- Date: Tue, 30 Sep 2008 15:20:46 +0200 From: "Erik Huelsmann" gmail.com> Subject: [j-devel] Quick task for volunteer To: "armedbear-j-devel< at >lists.sourceforge.net" lists.sourceforge.net> Message-ID: mail.gmail.com> Content-Type: text/plain; charset=UTF-8 Yesterday evening, I noticed our project page isn't XHTML compliant, even though it claims to be. Someone looking for a quick contribution: this is the ideal 15 minute or half hour job you can contribute! Thanks for any takers! Bye, Erik. PS: When you're taking, please mail the list so others know not to bite. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Closure.java cleanups, round 6

Ville Voutilainen — 2008-10-01T16:14:04

Very simple for-loop modification to use new for loop syntax where applicable. No regressions for ansi tests. Net win 4 lines. :) The actual net win is mostly in the row lengths of the changed lines, and clarity. processVariables contains copy-paste code for optionalParameters and keywordParameters, so there's a need for a tiny helper function. -VJV- P.S. I'll try to look at the more vexing issues, but I have no master plan where to look for the compiler lambda-block issue. Nor have I any master plan how to make compiler work with special decls. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ armedbear-j-devel mailing list armedbear-j-devel< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/armedbear-j-devel

Quick task for volunteer

Erik Huelsmann — 2008-09-30T13:20:46

Yesterday evening, I noticed our project page isn't XHTML compliant, even though it claims to be. Someone looking for a quick contribution: this is the ideal 15 minute or half hour job you can contribute! Thanks for any takers! Bye, Erik. PS: When you're taking, please mail the list so others know not to bite. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Closure.java cleanups, round 5

Ville Voutilainen — 2008-09-29T21:09:46

Yet another cleanup, consolidate the try blocks in 1-8 parameter execute functions into bindParameters, rename helper to bindParametersAndExecute. Tests still run, no regressions, net win 47 lines. Next cleanup patch for Closure.java will most be using the new for loop where applicable. I also intend to look at SpecialOperators.java and see if progn invocations can be used there in place of evaluating loops, if there's noone else doing that in the meantime..? That's a very minor thing, only two eval loops in there. -VJV- ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ armedbear-j-devel mailing list armedbear-j-devel< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/armedbear-j-devel

MOP tests

Ville Voutilainen — 2008-09-28T18:05:59

I found these quite recently: http://common-lisp.net/project/closer/downloads.html I'll run them next week, interesting to see what the results are. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Knowledge machine..

2008-09-28T08:57:50

------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ armedbear-j-devel mailing list armedbear-j-devel< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/armedbear-j-devel

RETURN handled differently in interpreted and compiledcode

Robert Dodier — 2008-09-27T23:04:49

Hello, See below for foobar.lisp. Interpreted: (foobar) => HELLO 1 NIL Compiled: (foobar) => HELLO 1 HELLO 2 T I'm no expert on the CL spec so I don't know which is correct. But I think Maxima expects the behavior shown by interpreted code here. Thanks for your attention to this. Robert Dodier PS. $ cat foobar.lisp (defun blurf () nil) (defun foobar () (prog (x y z) (declare (ignore x y z)) ((lambda (a b) (declare (ignore a b)) (format t "HELLO 1~%") (cond ((not (blurf)) (return nil)))) nil nil) (format t "HELLO 2~%") (return t))) ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

New patch for 'build.xml'

Mark Evenson — 2008-09-27T14:45:36

[Grrr! Somehow I lost a long reply to Ville about the build system: here's a shortened reply, but with a new conolidated patch.] 'build.xml' is not connected in any way to the 'autoconf' (i.e. the start a build by running './configure') way of building ABCL. There are actually three different ways of building ABCL, which I started to document in the lost post, but will write up again, and put on the [Google Code abcl-dynanmic-install wiki][1]. [1]: http://code.google.com/p/abcl-dynamic-install/wiki/BuildingFromSource Attached is a new patch to common-lisp.net code that is mostly complete from the perspective of building ABCL. The building J part of the patch is also more complete, but still doesn't make a satisfactory 'J' distribution that can be installed outside of the tree. Among other things, this patch creates a top-level 'abcl' wrapper, as Ville requested, on UNIX via the 'abcl.wrapper' target, i.e. unix$ ant abcl.wrapper will create a top-level 'abcl' script. Mark panix.com>

Closure.java cleanups, round 4

Ville Voutilainen — 2008-09-24T19:27:10

So, I haven't found yet anything on the compiler specials side, nor on the macrolet/symbol-macrolet side. I created a further patch for Closure.java in order to do at least something useful. This patch removes duplication from array-execute invocation by introducing a vararg helper. I also removed the "magic" numbers from the 1-8 parameter execute functions, because those blocks are all inside if (minArgs == x) where x is 1-8. Then we gave the same parameter to the previously introduced bindParameters helper. I changed this a bit so that we pass minArgs instead of hard-coded 1-8. Once again, this patch deletes more code than it adds, and it introduces no regression according to the ansi tests. Pretty straightforward, I think. There's more. :) After this patch, all evaluation blocks in the 1-8 parameter execute functions are identical: bindParameters(minArgs, ext, thread); try { return progn(body, ext, thread); } finally { thread.lastSpecialBinding = lastSpecialBinding; } This of course means that further (trivial) cleanup is yet again possible. We could create a separate helper again, but then again we can just add this bit of code to the end of bindParameters and call it bindParametersAndEval. We do need to pass in the lastSpecialBinding, though. I'd say there's the next patch I'll cook if I cannot find anything more important/interesting. The patch is quite trivial so it can be done even if something more pressing arises. -VJV- ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ armedbear-j-devel mailing list armedbear-j-devel< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/armedbear-j-devel

slime, local variables

Alan Ruttenberg — 2008-09-23T21:42:20

Embedding ABCL in a Java application

Alessio Stalla — 2008-09-23T18:44:08

Hello everyone. I'm doing some experiments on embedding ABCL in a Java application. Depending on the results, I may give a talk about it to a bunch of Java developers :) But I have a simple question. Suppose I need to package some additional Lisp files together with ABCL (for example, to offer a nicer Lisp API to my Java application - I'm not talking of a complete Lisp library here). How is the preferred way of doing it? The Java way would be to store these files in a Jar file, and load them as classpath resources. Indeed by looking at ABCL's source I see that Load.java has the capability to load code from any stream, not just files, but those methods are private (probably for a reason ;-) The Lisp way instead should be to load them into ABCL, dump an image and package ABCL + the image into the application. The problems with this approach are 1) I don't know how to do it ;) 2) from the point of view of a Java developer, depending on a file outside a jar is not nice or at least it's not the common behaviour of Java libraries... Problem 1) is easily solved, and is one of the reasons behind this post; problem 2) might not be a real problem, and an acceptable solution is not hard to find anyway. So, what do you think? Any suggestions? Thanks in advance, Alessio Stalla ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/