gmane.comp.programming.swig

Compilation errors when using Perl 5.8 headers,to upgrade the module from 5.6.1

Thomas Theakanath — 2008-08-21T00:46:48

------------------------------------------------------------------------- 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=/_______________________________________________ Swig-user mailing list Swig-user< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/swig-user

Re: Can't find jni files when using packages

Jim — 2008-08-20T11:14:42

Thank you for the tips. I'm a complete novice wrt swig. I'm trying to interface a bigger library and this is a made up sample of what I'm trying to do. I was using a sample interface file from the documentation rather blindly, I'm afraid, without understanding what it was doing. Thanks for the guidance. Jim. ------------------------------------------------------------------------- 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=/

Re: Can't find jni files when using packages

Solomon Gibbs — 2008-08-20T02:41:14

I'd suggest using only one %module directive per interface file. It appears to me that only the "%module book" directive is taking effect, because you specified "%module document" with a lower case 'd', but the compiler is trying to find "Document" with an upper case 'D'. Probably Document.java was produced by swig wrapping the Document type instead of by the "%module document" directive as you may have expected. I wrap static functions, constructors, etc. in a module names something like "libfay" and mostly allow swig to name the types wrappers. A great deal depends on how your code is structured. I usually also write a custom interface file instead of trying to pull in the original header. YMMV. ------------------------------------------------------------------------- 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=/

Re: Can't find jni files when using packages

Jim — 2008-08-19T13:22:43

So am I to believe that is problem is unsolvable? How does anyone use swig and java? Are there any canned examples anywhere? The only mention of packages in the documentation I've been able to locate is this brief statement: "The generated Java classes can be placed into a Java package by using the -package commandline option. This is often combined with the -outdir to specify a package directory for generating the Java files." Which I did and left me with uncompilable java files. Can anyone point me to a better place to ask this question? Thanks, Jim. ------------------------------------------------------------------------- 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=/

Re: array parameters and pinning in c#

Antti Karanta — 2008-08-19T11:15:17

On Tue, 19 Aug 2008 01:30:14 +0300, William S Fulton fultondesigns.co.uk> wrote: This is arguably so, but I don't think it helps here. To illustrate, here's the generated code in the module class: $prestuff try { $imcall } finally { $poststuff } Now, if a scope is closed in $poststuff, the last closing scope actually closes the finally block. There may be some code after it, which then falls outside the scope of the finally block. So, the scopes opened in $prestuff above should be closed *after* the finally block above. As to what would be the best way to achieve this, I don't know. I can take a shot at it, these are rather easy to write. YMMV. I don't think you can say it is universally the best way to do it. Pinning makes the work of the garbage collector more difficult, so e.g. in cases where the called native function makes calls back to the managed side via reverse p/invoke and those calls generate a lot of garbage, it might be better to use the default marshaling functionality (which copies the arrays). Another consideration is the size of the arrays passed. So, I think there need to be (at least) two sets of array typemaps to choose from: ones that fix the given array in memory and pass in a pointer to that, and another that uses the default p/invoke mechanism. Swig might default to one or the other, but allow the user to explicitly %apply either one. I can probably contribute both of these, given that using scopes becomes possible. Won't work, see above. The $post code needs to be inserted in the finally block, but the closing of scopes after it. But maybe a flag to indicate that the code in "pre" opens a scope that needs to be closed after the finally block has been closed? Then swig would just count how many params have this flag and insert corresponding number of closing braces after the finally block? It could be named e.g. "beginsscope", possible values being true/false. Or possibly it would take a number indicating how many scopes were opened, as one may want to begin several scopes in one "pre"? ::Antti:: ------------------------------------------------------------------------- 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=/

Re: array parameters and pinning in c#

William S Fulton — 2008-08-18T22:30:14

We do need some good array typemaps and so far your attempts are the best I've noticed. The pre and post attributes weren't designed with adding scope in mind. However, it doesn't seem an unreasonable thing to do. The problem is the order of the pre and post code is the same, when arguably the post code could be in reverse order to the pre code. I'd certainly be willing to think about adding in a feature to support this type of scoping. It would also allow one to add a 'using' block in as well as a 'fixed' block. If you are willing to develop and contribute these array typemaps and you think this is the best approach (most efficient) for marshalling arrays, I'll think of a way to do this, something like a new flag to indicate the post code contains a closing brace. William ------------------------------------------------------------------------- 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=/

Re: illegal Cast to array type

William S Fulton — 2008-08-18T21:57:21

I think this is one for the bug tracker as SWIG should generate code that will compile. As a workaround use the following: typedef unsigned char UCHAR; typedef const UCHAR CONSTTEST[6]; extern int say_hello(char * str); extern void test_func(CONSTTEST * arg1); William ------------------------------------------------------------------------- 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=/

Segmentation Faults with PHP bindings

Christopher Schwardt — 2008-08-18T10:41:34

Can't find jni files when using packages

Jim — 2008-08-17T15:44:25

I'm trying to figure out how to actually develop java library from c++ code. I've put a couple of simple classes together and done the following: swig -c++ -package com.fay.test -outdir src/com/fay/test -java demo.i gcc -I ${JAVA_HOME}/include -I ${JAVA_HOME}/include/linux -c demo_wrap.cxx ld -G demo_wrap.o -o libdemo.so Which worked fine. My interface looks like: %module book %{ #include "book.h" %} /* Let's just grab the original header file here */ %include "book.h" %module document %{ #include "document.h" %} Book has a reference to the Document class. When I attempt to compile the java files I get: jim< at >chinaberry:~/c++/src/com/fay/test$ javac *.java Document.java:31: cannot find symbol symbol : method delete_Document(long) location: class com.fay.test.bookJNI bookJNI.delete_Document(swigCPtr); ^ Document.java:37: cannot find symbol symbol : method new_Document() location: class com.fay.test.bookJNI this(bookJNI.new_Document(), true); ^ Document.java:41: cannot find symbol symbol : method Document_document_set(long,com.fay.test.Document,long) location: class com.fay.test.bookJNI bookJNI.Document_document_set(swigCPtr, this, SWIGTYPE_p_string.getCPtr(value)); ^ Document.java:45: cannot find symbol symbol : method Document_document_get(long,com.fay.test.Document) location: class com.fay.test.bookJNI return new SWIGTYPE_p_string(bookJNI.Document_document_get(swigCPtr, this), true); That's after I cd'ed to the source directory. The same thing happens if I do: javac src/com/fay/test/*.java So how do I compile? Environment: Ubunty 8.04, java package: sun-java6-jdk , 2.6.24-19 kernel, Thanks, Jim. ------------------------------------------------------------------------- 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=/

Re: C linked list to Python list/iterator

Viktor Hornak — 2008-08-17T13:55:49

Re: C linked list to Python list/iterator

Willi Richert — 2008-08-16T11:05:02

Re: boost shared_ptr incorrect deletion problem

Naveen Khanna — 2008-08-16T05:01:42

C linked list to Python list/iterator

Viktor Hornak — 2008-08-15T22:21:53

boost shared_ptr incorrect deletion problem

Naveen Khanna — 2008-08-15T21:32:57

array parameters and pinning in c#

Antti Karanta — 2008-08-15T06:22:17

Hi! Continuing my adventures w/ c# as the target language, I noticed that there were no typemaps for wrapping arrays (or did I just miss them?). No matter, should be easy enough to do. It seems p/invoke makes a copy of the array being passed, so I thought it would be more efficient to pin a pointer to the array and pass in that. So I did something like this: %typemap(ctype) int *INPUT_ARRAY, int &IN_ARRAY "int*" %typemap(imtype) int *INPUT_ARRAY, int &IN_ARRAY "IntPtr" %typemap(cstype) int *INPUT_ARRAY, int &IN_ARRAY "int[]" %typemap(csin, pre= " Debug.Assert( $csinput != null && $csinput.Length > 0 ) ;\n" " fixed ( int* csArrPtr$csinput = $csinput ) {\n", post="} // end fixed csArrPtr$csinput") int *INPUT_NARRAY, int &IN_NARRAY "(IntPtr)csArrPtr$csinput" %typemap(in) int *INPUT_ARRAY, int &INPUT_ARRAY "$1 = $input ;" %typemap(freearg) int *INPUT_ARRAY, int &INPUT_ARRAY "" %typemap(argout) int *INPUT_ARRAY, int &INPUT_ARRAY "" There is a slight problem, though - post attribute of csin typemap doesn't seem to be the right place for closing the scope of the fixed statement. The stuff there gets inserted into the finally block after the imcall and obviously closes that block. In some cases it doesn't matter, in some it does. So, has anybody figured out how to pass pinned arrays in c# using swig? ::Antti:: ------------------------------------------------------------------------- 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=/

illegal Cast to array type

Terence Kim — 2008-08-14T18:42:14

Re: handling comples macros

Mark Gossage — 2008-08-14T00:41:30

Hello Choonghwan, I have never seen such strange macros before, but after checking with my compiler, I must agree they do work. Now wrappering them: I found that this works quite well. ======================= %module demo; %constant int MACRO1; %constant int MACRO2; %include "header.h" ====================== Note: you can use either %constant int or const int. Both work, but they register the variable in different ways. If SWIG complains on redefinitions, try ======================= %module demo; %constant int MACRO1; %constant int MACRO2; %ignore MACRO1 %ignore MACRO2 %include "header.h" ====================== The ordering matters, you tell SWIG that MACRO1 is just an const integer (which it will believe), you dont need to give it a value. Then tell it to ignore all other information about it. This I think is the best way. Regards, Mark Send instant messages to your online friends http://uk.messenger.yahoo.com ------------------------------------------------------------------------- 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=/

Re: 64-bit Segmentation Fault with Swig 1.3.36

Brian Bird — 2008-08-13T14:56:12

Thanks! That seems to work perfectly. Now I just need to look into prototyping to find out *why* that works! :) ** to different Swig list to below) result tell PyObject SWIG_fail; fail; to the Python? work to be && and the segmentation believe size_t. ------------------------------------------------------------------------ great the ------------------------------------------------------------------------- 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=/

Re: Error in generated python wrapper code for statictemplate member function

Naveen Khanna — 2008-08-13T14:27:25

Re: 64-bit Segmentation Fault with Swig 1.3.36

Matthew O'Meara — 2008-08-13T13:28:54

Try using this interface file-- /* file /tmp/example.i */ %module example %{ extern char *test(char *, char *); %} extern char *test(char *, char *); Best, Matt On Wed, Aug 13, 2008 at 9:13 AM, Brian Bird securetrading.com> wrote: ------------------------------------------------------------------------- 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=/

Re: 64-bit Segmentation Fault with Swig 1.3.36

Josh Cherry — 2008-08-13T13:26:17

On Wed, 13 Aug 2008, Brian Bird wrote: Clearly the compiler believes that test() returns an integer. That might be because it is not prototyped. Josh ------------------------------------------------------------------------- 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=/