On Wed, 13 Nov 2013 08:15:48 -0500 Jim Jagielski <j...@jagunet.com> wrote:
> > On Nov 13, 2013, at 7:10 AM, Graham Leggett <minf...@sharp.fm> wrote: > > > On 13 Nov 2013, at 2:04 PM, Jim Jagielski <j...@jagunet.com> wrote: > > > >> It doesn't look like one to me... :/ > > > > One thing it does is make the return codes more targeted. Where > > previously any kind of failure would have returned 500 Internal > > Server Error[1] now we might return more specific codes based on > > the APR code returned. External code should remain unaffected. > > > > The question becomes is a subtle change to some return codes > > backportable or not? > > > > Well, we do, at times, adjust return codes to return the "right" > one (or the "better" one), during point/patch releases... > > But I'm not really pushing one way or another (yet) on this... My question (and I need to look at this in more detail, which means not today...) is whether this changes the required behavior by third party modules. Would all of our 2.4.6 modules continue to run correctly after the core change? If the answer is no, that means third party modules also require patching, which means this -is- an breaking API change.