On 9/12/07, Karl Fogel <kfogel@red-bean.com> wrote:
> "Garrett Rooney" <rooneg@electricjellyfish.net> writes:
> > FWIW, we have had various things in the SWIG bindings break as a
> > result of that kind of change in the past (I know, I'm the one who
> > broke them ;-), although it's usually the kind of thing where all of a
> > sudden some type mappings stop working, someone notices and fixes the
> > mappings. I don't think we've ever made any conscious decision to
> > support parameter names as some sort of firm API though, although I
> > could be mistaken there.
>
> Breaking the bindings is acceptable (though inconvenient), since they
> can just be fixed. Breaking the bindings *API* would be the problem...
Exactly. This is the kind of problem that introduces a temporary bug
that we then fix, at least in all cases I've encountered. Unless
there's some way SWIG leaks parameter names into the API (maybe for
named arguments in python?) we should be fine.
-garrett
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Sep 12 09:36:58 2007