|
From: Michael I. <mic...@ra...> - 2016-05-06 16:31:46
|
Hi Tom, Good point about making a clean break. I’ve committed the patches as originally submitted and updated any necessary calls in OpenBTS / SMQueue / SIPAuthServe. Thanks again for the submission! -Michael > On May 5, 2016, at 8:29 PM, Tom Tsou <tom...@et...> wrote: > > Hey Michael, > > On Thu, May 5, 2016 at 2:49 AM, Michael Iedema > <mic...@ra...> wrote: >> Could we not implement this in another way? Most of the altered calls in transceiver pass in sizeof(buf) for length. I suggest we keep the existing signature and default it to calculate a logical length in this way. Then we add the signature you’ve provided to improve the length specification when needed. > > In general, buffer pointers should not be passed without a size > specifier. Breaking the interface was explicit to that point. > > Is there a reason we can't fix the dependent read() calls? > > If we must, we can keep, but deprecate, the pointer only interface for > compatibility as in this patch; in the long run, use of that interface > really should go away. > > -TT > <0001-common-Add-previous-UDP-socket-interface-with-deprec.patch> |