Accepting request 226429 from home:kwk:branches:devel:tools:building

- Update to 3.0.0
  - This is a major new release focusing primarily on C++
    improvements.
  - C++11 support added. Please see documentation for details of
    supported features:
    http://www.swig.org/Doc3.0/CPlusPlus11.html
  - Nested class support added. This has been taken full advantage
    of in Java and C#. Other languages can use the nested classes,
    but require further work for a more natural integration into
    the target language.
  - We urge folk knowledgeable in the other target languages to
    step forward and help with this effort.
  - Lua: improved metatables and support for %nspace.
  - Go 1.3 support added.
  - Python import improvements including relative imports.
  - Python 3.3 support completed.
  - Perl director support added.
  - C# .NET 2 support is now the minimum. Generated using statements
    are replaced by fully qualified names.
  - Bug fixes and improvements to the following languages:
    C#, Go, Guile, Java, Lua, Perl, PHP, Python, Octave, R, Ruby, Tcl
  - Various other bug fixes and improvements affecting all
    languages.
  - Note that this release contains some backwards incompatible
    changes in some languages.
  - Full detailed release notes are in the changes file.
- drop 0001-Fix-Ruby-tracking-code-to-use-C-hash.patch,
  ruby-1.8.6-newobject.patch, and swig-fix_ocaml_type.patch.
  All upstreamed.

OBS-URL: https://build.opensuse.org/request/show/226429
OBS-URL: https://build.opensuse.org/package/show/devel:tools:building/swig?expand=0&rev=63
This commit is contained in:
Klaus Kämpf 2014-03-17 13:58:49 +00:00 committed by Git OBS Bridge
parent 66dae63279
commit 6f604dfc1b
7 changed files with 38 additions and 256 deletions

View File

@ -1,204 +0,0 @@
From 84d26240922e5f5b4c0365b0348e856c3c6aaf84 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Klaus=20K=C3=A4mpf?= <kkaempf@suse.de>
Date: Thu, 3 Jan 2013 11:56:25 +0100
Subject: [PATCH] Fix Ruby tracking code to use C hash
Patch id #330
This is a patch to resolve bug #2034216.
The bug is that the tracking code uses a ruby hash and thus may
allocate objects (Bignum) while running the GC. This was tolerated in
1.8 but is invalid (raises an exception) in 1.9.
The patch uses a C hash (also used by ruby) instead.
---
Lib/ruby/rubytracking.swg | 129 ++++++++++++++++++---------------------------
1 file changed, 50 insertions(+), 79 deletions(-)
diff --git a/Lib/ruby/rubytracking.swg b/Lib/ruby/rubytracking.swg
index 0a36f4a..02e92ad 100644
--- a/Lib/ruby/rubytracking.swg
+++ b/Lib/ruby/rubytracking.swg
@@ -22,19 +22,19 @@ extern "C" {
# error sizeof(void*) is not the same as long or long long
#endif
-
-/* Global Ruby hash table to store Trackings from C/C++
+/* Global hash table to store Trackings from C/C++
structs to Ruby Objects.
*/
-static VALUE swig_ruby_trackings = Qnil;
+static st_table* swig_ruby_trackings = NULL;
+
+VALUE get_swig_trackings_count(ANYARGS) {
+ return SWIG2NUM(swig_ruby_trackings->num_entries);
+}
-/* Global variable that stores a reference to the ruby
- hash table delete function. */
-static ID swig_ruby_hash_delete;
-/* Setup a Ruby hash table to store Trackings */
+/* Setup a hash table to store Trackings */
SWIGRUNTIME void SWIG_RubyInitializeTrackings(void) {
- /* Create a ruby hash table to store Trackings from C++
+ /* Create a hash table to store Trackings from C++
objects to Ruby objects. */
/* Try to see if some other .so has already created a
@@ -43,87 +43,47 @@ SWIGRUNTIME void SWIG_RubyInitializeTrackings(void) {
This is done to allow multiple DSOs to share the same
tracking table.
*/
- ID trackings_id = rb_intern( "@__trackings__" );
+ VALUE trackings_value = Qnil;
+ /* change the variable name so that we can mix modules
+ compiled with older SWIG's */
+ ID trackings_id = rb_intern( "@__safetrackings__" );
VALUE verbose = rb_gv_get("VERBOSE");
rb_gv_set("VERBOSE", Qfalse);
- swig_ruby_trackings = rb_ivar_get( _mSWIG, trackings_id );
+ trackings_value = rb_ivar_get( _mSWIG, trackings_id );
rb_gv_set("VERBOSE", verbose);
- /* No, it hasn't. Create one ourselves */
- if ( swig_ruby_trackings == Qnil )
- {
- swig_ruby_trackings = rb_hash_new();
- rb_ivar_set( _mSWIG, trackings_id, swig_ruby_trackings );
- }
-
- /* Now store a reference to the hash table delete function
- so that we only have to look it up once.*/
- swig_ruby_hash_delete = rb_intern("delete");
-}
-
-/* Get a Ruby number to reference a pointer */
-SWIGRUNTIME VALUE SWIG_RubyPtrToReference(void* ptr) {
- /* We cast the pointer to an unsigned long
- and then store a reference to it using
- a Ruby number object. */
-
- /* Convert the pointer to a Ruby number */
- return SWIG2NUM(ptr);
-}
-
-/* Get a Ruby number to reference an object */
-SWIGRUNTIME VALUE SWIG_RubyObjectToReference(VALUE object) {
- /* We cast the object to an unsigned long
- and then store a reference to it using
- a Ruby number object. */
-
- /* Convert the Object to a Ruby number */
- return SWIG2NUM(object);
-}
-
-/* Get a Ruby object from a previously stored reference */
-SWIGRUNTIME VALUE SWIG_RubyReferenceToObject(VALUE reference) {
- /* The provided Ruby number object is a reference
- to the Ruby object we want.*/
-
- /* Convert the Ruby number to a Ruby object */
- return NUM2SWIG(reference);
+ /* The trick here is that we have to store the hash table
+ pointer in a Ruby variable. We do not want Ruby's GC to
+ treat this pointer as a Ruby object, so we convert it to
+ a Ruby numeric value. */
+ if (trackings_value == Qnil) {
+ /* No, it hasn't. Create one ourselves */
+ swig_ruby_trackings = st_init_numtable();
+ rb_ivar_set( _mSWIG, trackings_id, SWIG2NUM(swig_ruby_trackings) );
+ }
+ else {
+ swig_ruby_trackings = (st_table*)NUM2SWIG(trackings_value);
+ }
+
+ rb_define_virtual_variable("SWIG_TRACKINGS_COUNT", get_swig_trackings_count, NULL);
}
/* Add a Tracking from a C/C++ struct to a Ruby object */
SWIGRUNTIME void SWIG_RubyAddTracking(void* ptr, VALUE object) {
- /* In a Ruby hash table we store the pointer and
- the associated Ruby object. The trick here is
- that we cannot store the Ruby object directly - if
- we do then it cannot be garbage collected. So
- instead we typecast it as a unsigned long and
- convert it to a Ruby number object.*/
-
- /* Get a reference to the pointer as a Ruby number */
- VALUE key = SWIG_RubyPtrToReference(ptr);
-
- /* Get a reference to the Ruby object as a Ruby number */
- VALUE value = SWIG_RubyObjectToReference(object);
-
/* Store the mapping to the global hash table. */
- rb_hash_aset(swig_ruby_trackings, key, value);
+ st_insert(swig_ruby_trackings, (st_data_t)ptr, object);
}
/* Get the Ruby object that owns the specified C/C++ struct */
SWIGRUNTIME VALUE SWIG_RubyInstanceFor(void* ptr) {
- /* Get a reference to the pointer as a Ruby number */
- VALUE key = SWIG_RubyPtrToReference(ptr);
-
/* Now lookup the value stored in the global hash table */
- VALUE value = rb_hash_aref(swig_ruby_trackings, key);
-
- if (value == Qnil) {
- /* No object exists - return nil. */
- return Qnil;
+ VALUE value;
+
+ if (st_lookup(swig_ruby_trackings, (st_data_t)ptr, &value)) {
+ return value;
}
else {
- /* Convert this value to Ruby object */
- return SWIG_RubyReferenceToObject(value);
+ return Qnil;
}
}
@@ -132,12 +92,8 @@ SWIGRUNTIME VALUE SWIG_RubyInstanceFor(void* ptr) {
since the same memory address may be reused later to create
a new object. */
SWIGRUNTIME void SWIG_RubyRemoveTracking(void* ptr) {
- /* Get a reference to the pointer as a Ruby number */
- VALUE key = SWIG_RubyPtrToReference(ptr);
-
- /* Delete the object from the hash table by calling Ruby's
- do this we need to call the Hash.delete method.*/
- rb_funcall(swig_ruby_trackings, swig_ruby_hash_delete, 1, key);
+ /* Delete the object from the hash table */
+ st_delete(swig_ruby_trackings, (st_data_t *)&ptr, NULL);
}
/* This is a helper method that unlinks a Ruby object from its
@@ -147,10 +103,25 @@ SWIGRUNTIME void SWIG_RubyUnlinkObjects(void* ptr) {
VALUE object = SWIG_RubyInstanceFor(ptr);
if (object != Qnil) {
+ if (TYPE(object) != T_DATA)
+ abort();
DATA_PTR(object) = 0;
}
}
+/* This is a helper method that iterates over all the trackings
+ passing the C++ object pointer and its related Ruby object
+ to the passed callback function. */
+
+/* Proxy method to abstract the internal trackings datatype */
+static int _ruby_internal_iterate_callback(void* ptr, VALUE obj, void(*meth)(void* ptr, VALUE obj)) {
+ (*meth)(ptr, obj);
+ return ST_CONTINUE;
+}
+
+SWIGRUNTIME void SWIG_RubyIterateTrackings( void(*meth)(void* ptr, VALUE obj) ) {
+ st_foreach(swig_ruby_trackings, (int (*)(ANYARGS))&_ruby_internal_iterate_callback, (st_data_t)meth);
+}
#ifdef __cplusplus
}
--
1.7.10.4

View File

@ -1,10 +0,0 @@
diff -wruN -x '*~' -x '*.o' -x '*.a' -x '*.so' -x '*.so.[0-9]' -x autom4te.cache -x .deps -x .libs -x Makefile -x Makefile ../orig-swig-2.0.10/Examples/test-suite/ruby/newobject1_runme.rb ./Examples/test-suite/ruby/newobject1_runme.rb
--- ../orig-swig-2.0.10/Examples/test-suite/ruby/newobject1_runme.rb 2013-05-27 20:24:04.000000000 +0200
+++ ./Examples/test-suite/ruby/newobject1_runme.rb 2013-05-29 09:55:57.922485272 +0200
@@ -29,5 +29,5 @@
GC.stats if $VERBOSE
swig_assert( 'Foo.fooCount == 200', binding, "but is #{Foo.fooCount}" )
GC.start
-swig_assert( 'Foo.fooCount <= 2', binding, "but is #{Foo.fooCount}" )
+swig_assert( 'Foo.fooCount <= 3', binding, "but is #{Foo.fooCount}" )

View File

@ -1,3 +0,0 @@
version https://git-lfs.github.com/spec/v1
oid sha256:65e13f22a60cecd7279c59882ff8ebe1ffe34078e85c602821a541817a4317f7
size 5312394

3
swig-3.0.0.tar.gz Normal file
View File

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:99cab1055877d00280509f0e04c1fe76643ec3792e9b8e7efcc9aa6e247229df
size 5556319

View File

@ -1,24 +0,0 @@
From: Philipp Thomas <pth@suse.de>
Date: 2013-03-27 15:47:28+01:00
Subject: Fix return type of caml_array_length
caml_array_type changed to mlsize_t, so fix the declaration.
---
Lib/ocaml/ocamldec.swg | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Index: Lib/ocaml/ocamldec.swg
===================================================================
--- Lib/ocaml/ocamldec.swg.orig 2012-09-07 08:06:00.000000000 +0200
+++ Lib/ocaml/ocamldec.swg 2013-03-27 15:46:36.135652282 +0100
@@ -130,7 +130,7 @@ CAMLextern int64 Int64_val(caml_value_t
SWIGSTATIC CAML_VALUE caml_array_new( int n );
SWIGSTATIC void caml_array_set( CAML_VALUE arr, int n, CAML_VALUE item );
SWIGSTATIC CAML_VALUE caml_array_nth( CAML_VALUE arr, int n );
- SWIGSTATIC int caml_array_length( CAML_VALUE arr );
+ SWIGSTATIC mlsize_t caml_array_length( CAML_VALUE arr );
SWIGSTATIC CAML_VALUE caml_val_char( char c );
SWIGSTATIC CAML_VALUE caml_val_uchar( unsigned char c );

View File

@ -1,3 +1,37 @@
-------------------------------------------------------------------
Mon Mar 17 07:46:53 UTC 2014 - kkaempf@suse.com
- Update to 3.0.0
- This is a major new release focusing primarily on C++
improvements.
- C++11 support added. Please see documentation for details of
supported features:
http://www.swig.org/Doc3.0/CPlusPlus11.html
- Nested class support added. This has been taken full advantage
of in Java and C#. Other languages can use the nested classes,
but require further work for a more natural integration into
the target language.
- We urge folk knowledgeable in the other target languages to
step forward and help with this effort.
- Lua: improved metatables and support for %nspace.
- Go 1.3 support added.
- Python import improvements including relative imports.
- Python 3.3 support completed.
- Perl director support added.
- C# .NET 2 support is now the minimum. Generated using statements
are replaced by fully qualified names.
- Bug fixes and improvements to the following languages:
C#, Go, Guile, Java, Lua, Perl, PHP, Python, Octave, R, Ruby, Tcl
- Various other bug fixes and improvements affecting all
languages.
- Note that this release contains some backwards incompatible
changes in some languages.
- Full detailed release notes are in the changes file.
- drop 0001-Fix-Ruby-tracking-code-to-use-C-hash.patch,
ruby-1.8.6-newobject.patch, and swig-fix_ocaml_type.patch.
All upstreamed.
-------------------------------------------------------------------
Sun Feb 9 18:30:13 UTC 2014 - kkaempf@suse.com

View File

@ -17,7 +17,7 @@
Name: swig
Version: 2.0.12
Version: 3.0.0
Release: 0
Summary: Simplified Wrapper and Interface Generator
License: GPL-3.0+ and BSD-3-Clause
@ -25,12 +25,6 @@ Group: Development/Languages/C and C++
Url: http://www.swig.org/
Source: http://sourceforge.net/projects/swig/files/swig/%{name}-%{version}/%{name}-%{version}.tar.gz
Source1: %{name}.rpmlintrc
# Fix the return type in declaration of caml_array_length
Patch9: swig-fix_ocaml_type.patch
# Fix SWIG object tracking to use C hash, not Ruby Hash, kkaempf@suse.de
Patch11: 0001-Fix-Ruby-tracking-code-to-use-C-hash.patch
# Ruby 1.8.6 (SLE 10) differs in object tracking, kkaempf@suse.de
Patch12: ruby-1.8.6-newobject.patch
BuildRequires: autoconf
BuildRequires: automake
BuildRequires: boost-devel
@ -124,14 +118,6 @@ understandig SWIG usage.
%prep
%setup -q
%if 0%{?suse_version} == 1010
%patch12 -p1
%endif
# Ruby 1.9 for openSUSE 12.x
%if 0%{?suse_version} >= 1220
%patch11 -p1
%endif
%patch9
%build
%configure --disable-ccache