test against the enum scoping problem I had.
Ocaml.cxx: Accidentally reintroduced the enum scoping problem, so I added
an example that will fail if this is ever broken again.
I encountered the actual problem in avifile.h
git-svn-id: https://swig.svn.sourceforge.net/svnroot/swig/trunk/SWIG@5478 626c5289-ae23-0410-ae9c-e8d60b6d4f22
makedebugtop: include swig.cmo
Lib: factored out more common code, slightly reorganized class type.
added director define and exceptions.
std_string: length from the original ocaml string (no longer depends on
null termination)
ocamldec.swg/ocaml.swg: added caml_string_len
Examples/Makefile.in: quiet about checking out files.
Ocaml.cxx:
Fixed abstract director test case.
Include/exclude director.swg based on directors being enabled.
Final edits for 1.3.20, barring bugs being discovered.
git-svn-id: https://swig.svn.sourceforge.net/svnroot/swig/trunk/SWIG@5477 626c5289-ae23-0410-ae9c-e8d60b6d4f22
William Fulton and Dave were talking about regarding the template parameters
and (), and this solved a bunch of wierd little problems I was having.
git-svn-id: https://swig.svn.sourceforge.net/svnroot/swig/trunk/SWIG@5455 626c5289-ae23-0410-ae9c-e8d60b6d4f22
- Compliance with ISO/IEC 14882:1998(E) 17.4.3.1.2 -> eg double underscores removed
- Potential member variable and method name clashes remove by preceding with 'swig'
- consistent use of C++ booleans for the swig_disown flag across and within the modules that use this it in director code.
git-svn-id: https://swig.svn.sourceforge.net/svnroot/swig/trunk/SWIG@5140 626c5289-ae23-0410-ae9c-e8d60b6d4f22
because that name better reflects what the function's trying to accomplish
(test whether the object for which the constructor's being called is an
instance of the class, or an instance of one of its subclasses).
git-svn-id: https://swig.svn.sourceforge.net/svnroot/swig/trunk/SWIG@4722 626c5289-ae23-0410-ae9c-e8d60b6d4f22
I've moved completely away from the "value" type to "caml_value_t". Although
a bit painful, this will make things easier for me. I seem to be seeing
quite a few things that use the word value as a function, type, etc, and
wanted to get something that doesn't collide.
git-svn-id: https://swig.svn.sourceforge.net/svnroot/swig/trunk/SWIG@4593 626c5289-ae23-0410-ae9c-e8d60b6d4f22
Type registration. Correct support for casting. Proper type safety.
Changed:
ocaml.swg: Correct casts, declaration order for C
ocamldec.swg: Removed problematic forward declarations without array index.
git-svn-id: https://swig.svn.sourceforge.net/svnroot/swig/trunk/SWIG@4573 626c5289-ae23-0410-ae9c-e8d60b6d4f22
Strings <=> std::string by value
std::wstring accessible from Ocaml. The string example converts a multibyte
japanese EUC sequence to a single wchar_t sequence if you have the ja_JP.EUC-JP
locale, or similar.
Better handling of reference in types
Corrected problems with & * mismatch in type verifier.
Type verifier now really functional. No more type errors in places they
wouldn't be allowed in C++, unless you work at it.
Added argout_ref example for argout_ref.
Init code now effective (called from let _ = f_<module>_init ())
git-svn-id: https://swig.svn.sourceforge.net/svnroot/swig/trunk/SWIG@4412 626c5289-ae23-0410-ae9c-e8d60b6d4f22
simple scheme I use for templating to work (at least in the test cases
so far). I probably need to come up with a better scheme for associating
the class closure creator with the typemaps.
git-svn-id: https://swig.svn.sourceforge.net/svnroot/swig/trunk/SWIG@4323 626c5289-ae23-0410-ae9c-e8d60b6d4f22
John Lenz