On Fri, Feb 21, 2014 at 01:58:41PM -0500, Dmitri Pal wrote:
Are there implication in handling all of them the same new way?
I mean if we assume that by something being equal to 0 we disable it
, would it cause an undesirable semantic change in some cases?
What about the cases when there is no sense in something being
disabled how we will treat the 0 value then?
For bools it is even more scary. If you say that bool = 0 sets the
default this is a bug because IMO bool = 0 means bool = false and
this is an explicit setting.
Now it is broken but it is consistent. I am concerned that we are
heading into the case when the options will be broken on case by
case basis.
What we agreed on with Sumit on IRC was to:
a) change the dp_copy_options function to only copy the current
values, never defaults.
b) for cases where we need to copy defaults (initialization of
subdomain providers for example), introduce a new function.
This way, both cases would be served by specific functions.