On Sunday, June 04, 2017 05:56:15 Jack Stouffer via Digitalmars-d-announce wrote: > On Saturday, 3 June 2017 at 06:09:21 UTC, Jonathan M Davis wrote: > > On Saturday, June 03, 2017 02:00:13 Jack Stouffer via > > > > Digitalmars-d-announce wrote: > >> I recommend a longer deprecation cycle than usual for this, as > >> this will break many legacy libraries that don't get > >> maintained often. A period of two years sounds about right. > > > > For Phobos, that _is_ the normal length of the deprecation > > cycle. For the language itself, I don't think that it's > > anywhere near as consistent, but I've gotten the impression > > that deprecations in the language usually stick around for > > quite awhile, but I haven't exactly tracked it closely, so I > > don't know. > > > > - Jonathan M Davis > > All of the recent Phobos deprecations have been a year. 18 months > at most.
If you think that, I think that you misunderstand how the Phobos deprecation process works. When a symbol is deprecated, it's marked in the documentation with the year-month that it will be removed from the documentation (usually about one year from the point that it's deprecated). Once that year has passed, the documentation is removed from Phobos, and instead, it's marked with a non-ddoc comment stating that the symbol is explicitly undocumented and that it will be removed at year-month where that's usually a year from when the symbol is removed from the documentation. Once that second date arrives, the symbol is completely removed. So, the whole deprecation cycle is approximately two years, and if anything, it sometimes takes a bit longer, because I'm sometimes slow to move the symbol along to the next stage. I suspect that what's confused you is that when the symbol is deprecated, it states in the documentation that the symbol will be removed at year-month and does not say anything about the fact that that removal date is when it will be removed from the documentation, not when it will be fully removed from the library. - Jonathan M Davis
