digitalmars.D - =?UTF-8?Q?Re=3A_std=2Eexperimental_=E2=80=93_DConf=3F?=
- Jonathan M Davis via Digitalmars-d (12/22) May 29 2014 It _should_, but Java had the ...
On Thu, 29 May 2014 19:46:38 -0700 Walter Bright via Digitalmars-d <digitalmars-d puremagic.com> wrote:On 5/29/2014 6:28 PM, Jonathan M Davis via Digitalmars-d wrote:If we make it clear up front that our policy is that nothing will stay in std.experimental permanently (at some point, either it gets moved into std in one form or another or it gets removed entirely), we explicitly state that the API in std.experimental can (and likely will) change, and we explicitly state that we will _not_ support the code in std.expermental long term unless/until it gets moved into std, then anyone who relies on std.experimental never changing has no one to blame but themselves.
Doesn't "experimental" imply all of that already?
It _should_, but Java had the javax disaster where they ended up keeping it as-is rather than treating it as experimental in order to avoid breaking code. Spelling out experimental will likely help a lot with that (whereas just putting an x on the end like java did isn't particularly clear), but we should probably still be very explicit about it in order to make sure that we avoid issues like that. It really does need to be invalid (and understood as invalid) to complain about code being broken by changes in std.experimental. Still, one would _hope_ that the fact that it's labeled experimental would be enough to get that across. - Jonathan M Davis
May 29 2014