[orbitcpp-list] Re: any patch

[Phil D. <ph...@li...>]

Gordon Miller wrote:
> 
> 
> Phil,
> 
> Attached is the patch that I created according to your instructions.  It
> seems awfully large to be a completely clean build so there might be
> some executable code trapped in there.  If there is, I apologize.  I
> will attempt to do a cleaner job in the future.  I have to admit a
> little bit of ignorance in the ways of patch.
> 

The patch worked a treat - I've committed your code to CVS.
Could I ask you to do a fresh checkout from cvs and work from that? I
noticed that there was some old code in pyIDL (accept() methods that
were used in the old days when I used the visitor pattern for the idl
compiler rather than the template-method pattern). I've no idea how they
could have got there unless there's some dodgy CVS versioning going on
somewhere, and a fresh checkout should fix that.

I noticed that you cleaned up the pyIDL code a bit to fit on 80 char
width screen; Thanks for that - I get a bit carried away with long lines
sometimes. 

BTW, There were some build files in the patch, however this will be
pretty unavoidable until I fix the 'make maintainer-clean' target so
that it works and clears out all the build files properly. It's
currently getting mixed up and attempting to build the cpp files in the
test directory using the idl compiler that has already been 'cleaned'
and so doesn't work. Once that's fixed, you'll be able to do a 'make
maintainer-clean' before doing your diff and all will be well.

> I have tested this by running it against the CosEventComm.idl and it
> appears to generate the correct idl.  I also have an any.py test that
> works against the any.idl that is also in the tests directory.

Yep. I added the any.idl file to the EXTRA_DIST target in the tests
Makefile.am. This makes sure that the idl file is included in the
distribution when I do a 'make dist' to generate it.


> I've also discovered that in order for us to really treat Anys, I'm
> going to have to create an Any object.  My plan is to just go ahead and
> wrap the existing Any stuff that is in ORBit (as it appears this is the
> general plan).  I have actually implemented Any objects in the past, so
> I'm pretty confident in my abilities for this.  Do you have a link to
> the CORBA C++ language mapping specification so that I don't miss
> anything?

You should be able to get the C++ language mapping spec on 
http://www.omg.org/library/clangindx.html

N.B. Unfortunately ORBit-C++ doesn't support typecode yet either - if my
understanding is correct, we'll need that before a full implementation
of any is possible.

cheers,

Phil.