On Mar 30, 2018, at 12:53 PM, Alan DeKok <aland@deployingradius.com> wrote:
On Mar 29, 2018, at 5:04 PM, Adam Bishop <Adam.Bishop@jisc.ac.uk> wrote:
On 29 Mar 2018, at 19:12, Arran Cudbard-Bell <a.cudbardb@freeradius.org> wrote:
As discussed offline, there are some side effects which makes this difficult to do whilst maintaining backwards compatibility.
v3.2.x?
That may be the thing to do. It might also give us the opportunity to do some minor cleanups in v3, too.
We killed v3.1.x because it would confuse people. If we create v3.2.x from the v3.0.x branch the initial reason for killing v3.1.x would be invalid, and it would definitely confuse people who deployed v3.1.x. Lets not fragment the development effort further. v4.0.x is deployed in production environments today, the unlang keywords are now nearly fully async, and we have work arounds for xlat/tmpl expansions that involve yielding which means we can add yield/resume points to all the modules gradually over time (after the initial release of v4.0.x). -Arran