Nikita, On Mon, Jul 16, 2012 at 5:30 AM, Nikita Popov <nikita....@gmail.com> wrote:
> On Mon, Jul 16, 2012 at 8:04 AM, Galen Wright-Watson <ww.ga...@gmail.com> > wrote: > > What about an approach like PDO, where the password functions would > > generate errors by default, but could be configured to throw exceptions? > > The ugliest aspects of this idea are the requirement for another function > > (password_set_option?) and hidden state. On the plus side, if you stick > > with errors for now it could be implemented down the road and be fully > BC. > > Guys, the whole point of the password API was to create something > simple and easy to use. Adding error configuration etc would make the > whole thing moot. > > I really don't see the issue with just throwing warnings here, like > every single other function does. Sure, exceptions would be nice, but > exceptions would be nice *everywhere*. They aren't of particular > importance here. If you want to discuss moving the internal functions > to some new error model, please start a new thread. It is definitely > something worth discussing, but is rather off-topic here. Definitely agree. Unless the discussion is specific to this RFC, please start a new thread for any commentary on errors in core... Anthony