On Fri, 20 Apr 2007, Joe Swatosh wrote:
> Hi Dan,
> On 4/20/07, Daniel Rall <firstname.lastname@example.org> wrote:
> >On Thu, 19 Apr 2007, Joe Swatosh wrote:
> >> Hi Paul,
> >> On 4/18/07, email@example.com <firstname.lastname@example.org> wrote:
> >> >Author: pburba
> >> >Date: Wed Apr 18 18:01:05 2007
> >> >New Revision: 24648
> >> >
> >> >Log:
> >> >Move svn_mergeinfo__to_string to new svn-private header for mergeinfo
> >> >
> >> This caused a test failure in the Ruby bindings:
> >> test_merge_info_to_s(SvnCoreTest):
> >> NoMethodError: undefined method `mergeinfo__to_string' for
> >> D:/SVN/src-trunk/subversion/bindings/swig/ruby/svn/core.rb:643:in
> >> ./subversion\bindings\swig\ruby\test\test_core.rb:748:in
> >> `test_merge_info_to_s'
> >> Are the private functions intended to be available to the bindings
> >> implementations or only to subversion itself? If the bindings are
> >> supposed to be able to use them, we will probably have to create a
> >> proxy for this?
> >I think that the bindings should be able to use this routine
> >internally. Exposing, or at least publicizing it, outside of
> >Subversion is more dubious...
> Thanks Dan, it sounds like headers in the private directory should
> have swig proxies
> generated for them, too? If yes, I'll start trying to figure that out....
I suppose. But, this makes the APIs easily accessible outside of
Subversion, which is questionable. Because these APIs are private to
Subversion, there's no guarantee that they won't change in the future
(no API compat).
Received on Mon Apr 23 18:25:38 2007
- application/pgp-signature attachment: stored