adam-faulkner / jnaerator Goto Github PK

when building I get a bunch of warnings on my machine:

sources/com/ochafik/swing/syntaxcoloring/TextAreaDefaults.java:56: warning:
unmappable character for encoding UTF8
        {*/ // on est oblig� de cr�er un objet nouveau � chaque fois
                           ^
You may want to avoid non ASCII characters in source and comments.

I'm trying to generate JNA interface using JNAerator for a C library but
I've got a lot of exception like this one:

line 1:10 Failed to match any alternative with token [@4,10:10='5',<4>,1:10]
         File: null:1
        Input: ( ( FEA_INT ) 5 )
         Rule: ()* loopback of 1407:3: (op= ( '<' | '<=' | '>' | '>=' ) f=
shift
Expr )*
        Stack: [expression, topLevelExprList, topLevelExpr, assignmentExpr,
inlineCondExpr, logOrExpr, logAndExpr, bitOrExpr, xorExpr, bitAndExpr,
equalExpr, compareExpr, shiftExpr, addExpr, multExpr, castExpr, unaryExpr,
postfixExpr, baseExpression, expression, topLevelExprList, topLevelExpr,
assignmentExpr, inlineCondExpr, logOrExpr, logAndExpr, bitOrExpr, xorExpr,
bitAndExpr, equalExpr, compareExpr]

Find attached the 3 header files and the stlib.dll if you want to reproduce.

My command line is:
java -jar ./lib/jna/jnaerator-v0.6-b367.jar -library stlib -o ./src
C:\FEA\lib\include\dllshare.h C:\FEA\lib\include\feaerror.h
C:\FEA\lib\include\stlib.h C:\FEA\lib\win32\stlib.dll

Logging of preprocessed sources and macros defaults to off. For debugging
the output might help (the jnaerator developer at least), so a switch would
be good. I will attach a patch to provide this with -logMacros and
-logPreprocessedSources.

Hi,

I am using MSYS/MINGW to compile FFMPEG.

The limits.h file that comes with MinGW as the following preprocessor #if
(the comment is from limits.h too...)

/* TODO: Is this safe? I think it might just be testing the preprocessor,
 *       not the compiler itself... */
#if ('\x80' < 0)

JNAerator does like the #if:
D:\test\msys\mingw\include\limits.h:41:2: error: Bad token in expression:
'\x80'
D:\test\msys\mingw\include\limits.h:41:2: error: missing ) in expression
D:\test\msys\mingw\include\limits.h:41:8: warning: Unexpected nonwhite token
D:\test\msys\mingw\include\limits.h:41:8: warning: Unexpected nonwhite token
D:\test\msys\mingw\include\limits.h:41:7: warning: Unexpected nonwhite token

This issue seems minor and is not causing further issues down the line.

What steps will reproduce the problem?
1. Launch JNAeratorStudio 
2. Click JNAerate! in the menu

What is the expected output? What do you see instead?
It should map "public com.sun.jna.NativeLong longLongValue;" to "public 
long longLongValue;", but instead it maps it to "public 
com.sun.jna.NativeLong longLongValue;"


What version of the product are you using? On what operating system?
JNAeratorStudio 0.4 (build 244). On windows XP

Please provide any additional information below.
long long is always 64 bit, either in 32 bit or 64 bit procesor, at least 
with gnu compiler (I believe microsoft compiler doesn't support this type, 
but haven't tested that) 

{{{
extern void (*fptr)(int);
}}}
fptr is not jnaerated.

Globals generation + type conversion need an overhaul anyway, big work ahead...

So that library options can be used for these structs

Building in linux (english UTF8 locale) is resulting in warnings:
warning: unmappable character for encoding UTF8

This is due to special characters in comments and/or source code.
This may lead to strange results depending on the platform.

Patch included. (btw I noted that at least at one point a catch Throwable
is happening, catch Exception should be used whenever possible, or even
more specific exception classes. If logging and/or rethrowing is performed,
attach the root exception for easier debugging.)

When the expressions are simple enough to be understood, we should provide 
the macros and inline functions encountered during the parsing so that 
they're available from Java.

Possible implementation approaches : 
- add a "MACROS" / "INLINE" / "_" subclass to generated classes
- add macros and functions as simple methods to the regular top-level 
library interface, but tag them with an annotation that somehow indicates 
where to find the implementation (inner class ?) and let a modified JNA do 
the job of wiring them (slow, partly negates interest of inline).

Currently the tests called by "-test" assume to be in a fixed location.

Patch to make them runable from the jar built by the jnaerator project.

What steps will reproduce the problem?

The issue occurs when I try to build my maven project depending on
jnaerator-runtime-0.9.1 (and not using SNAPSHOT deps).

1. get jnaerator-runtime-0.9.1.pom from
http://jnaerator.sourceforge.net/maven/com/jnaerator/jnaerator-runtime/0.9.1/jna
erator-runtime-0.9.1.pom
2. it contains:
<parent>
<artifactId>jnaerator-parent</artifactId>
<groupId>com.jnaerator</groupId>
<version>1.0-SNAPSHOT</version>
<relativePath>..</relativePath>
</parent>
3. SNAPSHOT here is very bad. 

What is the expected output? What do you see instead?

It should depend on a released version such as 0.9.1.
Released versions should never depend on snapshots. If you decide to
restructure the jnaerator-parent/pom.xml and deploy it as 1.0-SNAPSHOT
again, it probably will brake my build as it will change the configuration
of the already released jnaerator-runtime-0.9.1

What version of the product are you using? On what operating system?
jnaerator-runtime-0.9.1
java 1.6

This enhancement request was taken from Comment 2 of issue #36 by 
[email protected], Oct 
01, 2009 :
An additional requirement would be to define the static constants inside the
generated classes when the #define is found inside a struct (ffmpeg makes a lot 
of
use of this pattern).

It would require a change of the parser to identify the end of each element, 
and a change of the 
generator to match start/end offsets with #defines.

What steps will reproduce the problem?
1. mvn clean com.jnaerator:maven-jnaerator-plugin:jnaerate install on any
maven project with the plugin declared.

What is the expected output? What do you see instead?
The build aborts before compiling, saying "JNAeration completed !". I
expect compilation to happen.

What version of the product are you using? On what operating system?
version 0.92 on 64bit Ubuntu 9.10 

Please provide any additional information below.
The line causing the problem is
http://jnaerator.googlecode.com/svn/trunk/jnaerator/src/main/java/com/ochafik/la
ng/jnaerator/JNAerator.java
line 599. There are a few System.exit()s in that file.

What steps will reproduce the problem?
Generating this:
{{{
@interface A
-(id)f:(id)arg;
@end
}}}
gives :
{{{
package test;
public abstract class A {
    ...
    public abstract org.rococoa.NSObject f(org.rococoa.NSObject arg);
    ...
}
}}}

What is the expected output? What do you see instead?
We'd expect the arg to be org.rococoa.ID

What steps will reproduce the problem?
1.run jnaerator on c/c++ header file that doesn't have ".h" extension
2.
3.

What is the expected output? What do you see instead?
Parsing of the file. Generated output. Without extension I receive this:
"Now parsing 0 text blocks"

If i copy the file to contain an extension then I receive this:
"Now parsing 27 text blocks"

What version of the product are you using? On what operating system?
jnaerator-v0.9-b614.jar ,MacOSX 10.5.8

Please provide any additional information below.
I'm am trying to wrap a third party lib using jnaerator. The library is 
composed of many header 
files and many sub libraries. The issue I have is that the headers don't 
contain the ".h" suffix in 
the name. In fact, they have no suffix. Is it possible to add a flag/option to 
parse all files in a 
directory no matter the suffix?

- need to look at source file type to know if symbols are C++-mangled by 
default
- need to implement mangling schemes of VSC++2008 & GCC 4.x

What steps will reproduce the problem?
1.checkout source from subversion
2.type > mvn install
3.build fails with error

What is the expected output? What do you see instead?
completion of build

What version of the product are you using? On what operating system?
svn revision 892

Please provide any additional information below.
Attached the log output.

In sed version 4.1.5 there is no "-E" switch. I assume this is a typo and
it should read "-e".

What steps will reproduce the problem?
1. I donwload vlc source code from
http://www.videolan.org/vlc/download-sources.html (Latest VLC source code
tarball (1.0.1))
2. use command line : java -jar jnaerator.jar  -noComp -noJar -noAuto
-package vlc.java.binding libvlc.h

3. the api "libvlc_new" in libvlc.h was not generated into LibvlcLibrary.java 

What is the expected output? What do you see instead?


What version of the product are you using? On what operating system?
==I used Jnaerator build 621, winXP, JDK1.6.0_15

Please provide any additional information below.

I tried to build from source with the following JARs in classpath:

* antlr-3.1.1-runtime.jar
* binarysearch.jar ( http://ochafik.free.fr/blog/?p=106 )
* jalico-latest.jar
* jna.jar  ( 3.0.9 )
* junit.jar ( 4.4 )
* trove-2.0.4.jar


3rd Party source files missing:

* all files for 'slightly patched Anarres JCPP'


'com.ochafik' - source files missing:

* com.ochafik.babel.env.SchemeEnv
* com.ochafik.io.TextAreaOutputStream
* com.ochafik.math.graph.BinaryEdgeSet;
* com.ochafik.math.graph.impl.FastSparseBinaryEdgeSet;
* com.ochafik.swing.JDialogs
* com.ochafik.swing.tree.DefaultTreeNode
* com.ochafik.util.listenable.ListenableListModel


If you add these sources to SVN I will be able to create a Maven project
definition (pom.xml) which will enable everyone to build from source easily.

Hello,

I've got a lot of such exception while jnaerator does parse the header file:

Error in mangling of 'Sgetstripcurve(STRIPLETSTRUC*, short*, CHOICETYPE,
FEA_OUT
PUT*)' : java.lang.NullPointerException

Though the Sgetstripcurve function is generated anyway in the Java interface.

Where could it come from ?

What steps will reproduce the problem?
1. execute one of the examples: java -Xmx1000m -jar jnaerator-0.9.1.jar 
-framework 
Foundation -framework AppKit -jar appkit-foundation.jar
2.
3.

What is the expected output? What do you see instead?
The example should produce a working jar, instead the generation process fails 
with 33 errors 
(see complete log below).

What version of the product are you using? On what operating system?
JNAerator 0.9.1 on Mac OS X 10.6.1, Java 6.

Please provide any additional information below.
I attached the complete log as a txt file. Hope this helps!

What steps will reproduce the problem?
Take example code from [http://theocacao.com/document.page/510] :
{{{
@interface Movie : NSObject {

  NSString* title;
  NSString* studio;
  NSInteger yearReleased;
}

+ (id)movie;

@property (copy) NSString* title;
@property (copy) NSString* studio;
@property (assign) NSInteger yearReleased;
@property (readonly) NSString* summary;

@end
}}}

What is the expected output? What do you see instead?
We need to generate getters and/or setters for properties (readonly => no 
setter).

Currently the parser doesn't mind which language the input files were written 
into.

As a result, "class", which is a reserved keyword in C++, is not parsed as a 
valid identifier in C.

Likewise, it is hard to know whether a symbol should be C++-mangled (behaviour 
by default in 
*.cpp files) or not (behaviour by default in *.c files).

We should either provide a command-line switch or use the existing 
__cplusplus__ symbol to force 
a language.

What steps will reproduce the problem?
1.The resulting code produced includes lots of lines such as public class
SKYETEK_ADDRESS extends
com.ochafik.lang.jnaerator.runtime.Structure<SKYETEK_ADDRESS,
SKYETEK_ADDRESS.ByValue, SKYETEK_ADDRESS.ByReference>
2.I need to know where to find com.ochafik.lang.jnaerator
3.A similar problem occurs with generated lines:
@com.ochafik.lang.jnaerator.runtime.Mangling({"_Z20SkyeTek_CreateDeviceP5TCHARPP
14SKYETEK_DEVICE",
"?SkyeTek_CreateDevice@@YAW4SKYETEK_STATUS@@PA5TCHARPAPA14SKYETEK_DEVICE@Z"})

What is the expected output? What do you see instead?

I expect to have access to the runtime library so that these errors can be
removed and the project progress.


What version of the product are you using? On what operating system?
I am running NetBeans IDE 6.7.1 on Windows Vista. I believe I am running
0.9 b21


Please provide any additional information below.

Can provide the .h file if required

What steps will reproduce the problem?
1. svn co of trunk 
2. cd jnaerator
3. mvn install

What is the expected output? Build Successful

What do you see instead? listing of unmappable character locations in

DateUtils.java and FileUtils.java

src/main/java/com/ochafik/util/DateUtils.java:[25,31] unmappable character
for encoding UTF8

"récupérer" (it gets the column wrong by one: gedit sees it fine, which
seems odd).

src/main/java/com/ochafik/io/FileUtils.java:[130,40] unmappable character
for encoding UTF8

What version of the product are you using? 
Maven version: 2.0.9
Java version: 1.6.0_16 (sun)
OS name: "linux" version: "2.6.28-15-generic" arch: "i386" Family: "unix"

On what operating system?

Ubuntu 9.04

Please provide any additional information below.

I tried

javac -encoding utf8 src/main/java/com/ochafik/util/DateUtils.java

and same error. What encoding is being used? I tried MacRoman and
ISO-8859-1. I think this can be set in the pom.xml.

What steps will reproduce the problem?
1. Go to http://code.google.com/p/jnaerator/
2. Click on Java web start button
(http://ochafik.free.fr/Java/JNAeratorStudio.jnlp)
3. Program doesen't start

What is the expected output? What do you see instead?

com.sun.deploy.net.JARSigningException: Found unsigned entry in resource:
http://jnaerator.sourceforge.net/maven/com/jnaerator/jnaerator/0.9-SNAPSHOT/jnae
rator-0.9-SNAPSHOT.jar
    at
com.sun.javaws.security.SigningInfo.getCommonCodeSignersForJar(SigningInfo.java:
382)
    at com.sun.javaws.security.SigningInfo.check(SigningInfo.java:256)
    at
com.sun.javaws.LaunchDownload.checkSignedResourcesHelper(LaunchDownload.java:138
4)
    at
com.sun.javaws.LaunchDownload.checkSignedResources(LaunchDownload.java:1170)
    at com.sun.javaws.Launcher.prepareLaunchFile(Launcher.java:1020)
    at com.sun.javaws.Launcher.prepareToLaunch(Launcher.java:306)
    at com.sun.javaws.Launcher.prepareToLaunch(Launcher.java:183)
    at com.sun.javaws.Launcher.launch(Launcher.java:104)
    at com.sun.javaws.Main.launchApp(Main.java:403)
    at com.sun.javaws.Main.continueInSecureThread(Main.java:250)
    at com.sun.javaws.Main$1.run(Main.java:111)
    at java.lang.Thread.run(Thread.java:619)

What version of the product are you using? On what operating system?

Ubuntu 9.04
Sun Java 1.6
Firefox

Please provide any additional information below.

What steps will reproduce the problem?
the following structure 
struct SYSTEMTIME {
   char name[16];
};

is converted to 
public static class SYSTEMTIME extends com.sun.jna.Structure
    {
        public static class ByReference extends SYSTEMTIME implements 
com.sun.jna.Structure.ByReference {}

        public final com.sun.jna.ptr.ByteByReference name = new byte[16];

    }

What steps will reproduce the problem?
1. get the read-only code from svn as described
http://code.google.com/p/jnaerator/source/checkout
2. run mvn install
3. it tries to build a non-existing module maven-velocity-plugin. It fails
of course.
4. the pom.xml files include a SNAPSHOT version.

What is the expected output? What do you see instead?
3. It should not build the maven-velocity-plugin. Or the svn should
containg sources for it if it is really needed.
4. I do not want to build a snapshot. How can I get sources of a released
version (0.9.1)? What is its revision number? I can not see any tags in the
svn repo. It would be nice to see tags corresponding to the relesed versions.

What version of the product are you using? On what operating system?
trunk
java 1.6

What steps will reproduce the problem?

$ java -jar jnaerator-0.9.3-SNAPSHOT.jar -library Test Test.h -o . -v


What is the expected output? What do you see instead?

There should be a TestLibrary.java file.  What I see is:

$ ls -l
total 11824
-rw-r--r--@ 1 dspitzer  HOLLYWOOD\domain users      654 Nov 20 16:15 Test.h
-rw-r--r--  1 dspitzer  HOLLYWOOD\domain users   992493 Dec  3 10:58 Test.jar
-rw-r--r--  1 dspitzer  HOLLYWOOD\domain users        1 Dec  3 10:58 
_jnaerator.bridgesupport.h
-rw-r--r--  1 dspitzer  HOLLYWOOD\domain users      570 Dec  3 10:58 
_jnaerator.choices
-rw-r--r--  1 dspitzer  HOLLYWOOD\domain users      583 Dec  3 10:58 
_jnaerator.macros.cpp
-rw-r--r--  1 dspitzer  HOLLYWOOD\domain users      632 Dec  3 10:58 
_jnaerator.preprocessed.c
-rw-r--r--@ 1 dspitzer  HOLLYWOOD\domain users  5035045 Nov 26 10:32 
jnaerator-0.9.3-
SNAPSHOT.jar


What version of the product are you using? On what operating system?

jnaerator-0.9.3-SNAPSHOT.jar
java version "1.6.0_15"
Mac OS X 10.5.8


Please provide any additional information below.

One step to reproduce:
1. use Jnaerator to process any header file with hyphen in its name, like 
"test-me.h"

Expected is some compilable Java class name, like testMeLibrary or 
testmeLibrary or whatnot.

The observed result is "test-meLibrary" class which fails to be compiled. 
Same applies to the generated package names.

I'm running console jnaerator-0.9.1.jar on Windows Vista OS, with the 
following parameters:
java -Xmx1024M -jar jnaerator-0.9.1.jar -direct -entryClass MyLib -package 
org.my.lib mylib/*.h

My OpenSolaris box does not have a /usr/include/c++ directory, so that line
causes a null pointer.

As CLASS is only used by the class itself, we can replace all of its uses 
by calls to getCLASS(), for instance, that will create the CLASS instance 
the first time it is called.
This might speed up loading of large libraries, where hundreds of mutually-
referencing classes may be defined.

static APPKIT_EXTERN NSString constants do not get converted (reported by David 
Kocher)

The input:
===========================
SET ffmpg_src="C:\msys\home\xxx\ffmpeg-r19905-swscale-r29687"
SET ffmpg_build="%ffmpg_src%\build"
SET jre="C:\Program Files\Java\jre1.6.0_07\bin\java.exe"
SET includeVS="C:\Program Files\Microsoft Visual Studio 9.0\VC\include"
SET includeSDK="C:\Program Files\Microsoft SDKs\Windows\v6.0A\Include"
SET includeMGW="C:\msys\mingw\include"
SET includeLib1="C:\msys\mingw\include\schroedinger-1.0"
SET root="ffmpeg"
SET exec=%jre% -Xmx1024m -jar jnaerator-v0.9.1.jar -noComp -scanSymbols
-limitComments -direct -v -o . -root %root% -I %includeMGW% -I
%includeMGW%\sys -I %ffmpg_src% -I %ffmpg_src%\build -I %includeLib1%

%exec% -library AVCodec %ffmpg_src%\libavcodec
%ffmpg_build%\libavcodec\avcodec.dll


The output:
===========================
....
Generating roq_qcell.java
Generating motion_vect.java
Generating RoqContext.java
Generating RV34VLC.java
Generating SliceInfo.java
Generating RV34DecContext.java
JNAeration failed !
java.lang.NullPointerException
        at
com.ochafik.lang.jnaerator.Result.getFakePointerName(Result.java:170)
        at com.ochafik.lang.jnaerator.Result.findFakePointer(Result.java:149)
        at
com.ochafik.lang.jnaerator.TypeConversion.convertTypeToJNA(TypeConversion.java:9
83)
        at
com.ochafik.lang.jnaerator.DeclarationsConverter.convertVariablesDeclaration(Dec
larationsConverter.java:987)
        at
com.ochafik.lang.jnaerator.DeclarationsConverter.convertVariablesDeclaration(Dec
larationsConverter.java:1056)
        at
com.ochafik.lang.jnaerator.DeclarationsConverter.convertStruct(DeclarationsConve
rter.java:822)
        at
com.ochafik.lang.jnaerator.DeclarationsConverter.outputConvertedStruct(Declarati
onsConverter.java:896)
        at
com.ochafik.lang.jnaerator.DeclarationsConverter.convertStructs(DeclarationsConv
erter.java:977)
        at
com.ochafik.lang.jnaerator.JNAerator.generateLibraryFiles(JNAerator.java:1084)
        at
com.ochafik.lang.jnaerator.JNAerator.jnaerationCore(JNAerator.java:1251)
        at com.ochafik.lang.jnaerator.JNAerator.jnaerate(JNAerator.java:656)
        at com.ochafik.lang.jnaerator.JNAerator$1.finished(JNAerator.java:569)
        at
com.ochafik.lang.jnaerator.JNAeratorCommandLineArgs$ArgsParser.parse(JNAeratorCo
mmandLineArgs.java:127)
        at com.ochafik.lang.jnaerator.JNAerator.main(JNAerator.java:237)

Environment:
===========================
Windows XP SP2
Java 1.6.0._07-b06
jnaerator-v0.9.1.jar

{{{
int byref; // does not parse because of byref objective-c modifier not being 
ruled out here
}}}

Need to define scope variables in parser to enable/disable whole families of 
modifiers.

What steps will reproduce the problem?
{{{
typedef void (*f)();
void test(f* x);
struct S {
   f* pf;
};
}}}

What is the expected output? What do you see instead?
Should use Pointer or some typed structure, instead getting the raw function 
pointer (not the 
pointer to it !!)

Right now, Objective-C classes are jnaerated as java interfaces.
To make the jnaerated code more Rococoa-idiomatic, it is better to generate 
abstract classes (see example from David Kocher here : 
http://trac.cyberduck.ch/browser/branches/rococoa/source/ch/cyberduck/ui/co
coa/foundation/NSDate.java?rev=4698).

Logic of generation of static factory methods needs to be updated 
accordingly.

Initialization of the default include path is broken on linux, since an
unmodifieable list is created for the default include path. At startime
modifications fail.

patch:
Index: sources/com/ochafik/lang/jnaerator/JNAeratorConfigUtils.java
===================================================================
--- sources/com/ochafik/lang/jnaerator/JNAeratorConfigUtils.java    (revision 435)
+++ sources/com/ochafik/lang/jnaerator/JNAeratorConfigUtils.java    (working copy)
@@ -97,9 +97,11 @@
            list.add(".");
            DEFAULT_INCLUDE_PATH = Collections.unmodifiableList(list);
        } else {
-           DEFAULT_INCLUDE_PATH = Arrays.asList(".");
+           DEFAULT_INCLUDE_PATH = new ArrayList<String>();
+           DEFAULT_INCLUDE_PATH.add(".");
        }
        if (SystemUtils.isUnix()) {
+           DEFAULT_INCLUDE_PATH = new ArrayList<String>();
            DEFAULT_INCLUDE_PATH.add("/usr/include");

Hi,

It seems that the order of #define is not preserved in the generated java code.
The resulting "public static final int ..." statements are in random order.

It would be nice to preserve the original ordering in the generated java code.

Regards.

What steps will reproduce the problem?
1. Try to janearate glib.h on Linux Ubuntu 9.04
2. Set up Maven project with the following config values:
-I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include/linux
-I/usr/lib/gcc/i486-linux-gnu/4.3/include -library Gtk
/usr/include/glib-2.0/glib.h
3. Run the Maven JNAerator plugin

What is the expected output? What do you see instead?

Expected to get a JAR with GLib bindings. 
Instead get:

[INFO] Scanning for projects...
[INFO] ------------------------------------------------------------------------
[INFO] Building gtk
[INFO]    task-segment: [com.jnaerator:maven-jnaerator-plugin:jnaerate]
[INFO] ------------------------------------------------------------------------
[INFO] Preparing jnaerator:jnaerate
[INFO] No goals needed for project - skipping
[INFO] [jnaerator:jnaerate {execution: default-cli}]
os.arch = i386
/usr/include/glib-2.0/glib/gmessages.h:128:12: error: error in macro
parameters: ...
/usr/include/glib-2.0/glib/gmessages.h:133:37: error: error in macro
parameters: ...
/usr/include/glib-2.0/glib/gmessages.h:136:15: error: error in macro
parameters: ...
/usr/include/glib-2.0/glib/gmessages.h:139:14: error: error in macro
parameters: ...
/usr/include/glib-2.0/glib/gmessages.h:142:12: error: error in macro
parameters: ...
JNAeration completed !
/media/My Passport/Jacek/Dev/Scala/gtk4j/target/generated-sources/java

What version of the product are you using? On what operating system?

0.9.1, Ubuntu 9.04
Please provide any additional information below.

> What steps will reproduce the problem?

1. build some application that includes jnaerator and slf4j logging
2. try to provide logging through log4j using slf4j-log4j12 implementation
3. try to configure logging using log4j.properties

> What is the expected output? What do you see instead?

I expect the application reads log4j.properties and logs according to it.
The jnaerator hijacks all the logging and routes it through jdk14 logging
system. log4j.properties is ignored.

> What version of the product are you using? On what operating system?

0.9.1, but i see 0.9.2 does the same.

> Please provide any additional information below.

I would expect that jnaerator.jar comes only with its own classes, and it
gets other dependencies using its pom.xml. 

Including slf4j implementation together with the library jar misses the
main point of slf4j. It is meant that the libraries use only the API jar
(slf4j-api) and depend only on it. The application developer chooses which
implementation to use and include only that specific implementation.

Hi,

I am trying to run the latest (v0.9.1) jnaerator against the latest ffmpeg
Windows libraries.

The result are quite impressing as most dll are handled correctly, but
handling of avformat.dll is partial (stops midway...).

The command line I am running is:
java -jar jnaerator-0.9.1 FFMPEG.jnaerator

I have attached the FFMPEG.jnaerator file and the resulting out.errors.txt
file. I also attached out.txt which captures the stdout/stderr outputs.

Basically, jnaerator is reporting some errors in libavformat/avformat.h and
will not generate all the structs and methods found in this header.

The two offending lines in avformat.h are:
1) a member declaration in AVOutputFormat struct
    int (*interleave_packet)(struct AVFormatContext *, AVPacket *out,
                             AVPacket *in, int flush);
2) a method declaration
    int av_interleave_packet_per_dts(AVFormatContext *s, AVPacket *out,
                                 AVPacket *pkt, int flush);

Strangely, if in both cases I remove the "AVPacket *out, AVPacket *pkt,"
part, the jnaerator works fine and completes successfully.

I don't expect a quick fix ;-) as I am well aware that the information I am
providing might not be enough to understand and fix the issue.
I am ready to provide more information and try things out should you ask
for it...

Regards and thanks in advance.

http://msdn.microsoft.com/en-us/library/57a97k4e(VS.80).aspx

In other terms, 

void CopyBytes(char* dest, const char* source, size_t n, const int* intArray); 

no longer gets converted to :

void CopyBytes(ByteByReference destination, ByteByReference source,
NativeLong n, IntByReference intArray); 
void CopyBytes(ByteByReference destination, String source, NativeLong n,
int[] intArray);

We need to test that code based on JNAerated sources will keep on building 
with forecoming changes.

We could use the Java Compiler API + a set of C headers + associated Java 
files that refer to predefined library names that would be generated.

{{{ int* (*x)(); }}} isn't parsed right now.
This is a serious issue, related to the more general variables storage parsing 
limitations.

What steps will reproduce the problem?
JNAerate this :
{{{
struct S { int a; };
void f(S**);
}}}

What is the expected output? What do you see instead?
We're getting {{{ f(PointerByReference) }}}.
We should get {{{ f(S.ByReference[]) }}}

What steps will reproduce the problem?
1. Have C header file with 

    #define DEV_DEBUG ((uint64_t)1 << 8)

Same error with u_int64_t or int64_t

2. run jnaerator-0.9.1.jar
3. get error

What is the expected output? 
 -- proper parsing of a 64-bit unsigned long typecast 1 shifted 8 bits

What do you see instead?
 -- And error message 
    "Failed to convert define 'DEV_DEBUG => ((uint64_t)1 << 8):"

What version of the product are you using? 0.9.1 On what operating system?
Ubuntu 9.04

Attempt to run JNAerator on a Linux system to generate JNA interfaces for a
library.  Rather than using the normal configuration for C code and
considering /usr/include the base directory for finding C header files
(along with /usr/include/c++ and its subdirectories for the C++ STL) such
that "include <x/y.h>" locates the file /usr/include/x/y.h, JNAerator
requires that each and every directory from which a header file is included
by the library code be passed to it via -I.  This is a PITA since a user
should not have to know what headers from the standard library their target
library uses and where to find those headers.

I'm using 0.4-b408 as available on the downloads with an Ubuntu Linux 9.04
system, with all dependencies installed.


I'll delve into the source of JNAerator and see if I can't submit a patch
to fix this.