Deprecation policies

[John Marino]

On 3/10/2014 07:50, Alexey Dokuchaev wrote:
> On Sat, Mar 08, 2014 at 01:09:50PM +0100, Gerald Pfeifer wrote:
>> If anything, I think we need to consider becoming more aggressive:
>>
>> A port that fails to build on either of the latest two major release
>> branches for X months gets deprecated.
> 
> Fine by me, however, I'd added that whoever is deprecating it due to
> build breakage should try to unbreak it first: sometimes this is very
> easy to fix (like with net-p2p/microdc2).

I think it's pretty clear that this staging effort doubles as
housecleaning.  A lot of cruft is getting exposed, and a lot of missing
maintainers are getting discovered.  Personally I have no issue with
ports being on a very short lease.   One port gets resurrected out of
how many garbage candidates?  And the end result is that it was fixed?
I'm not seeing a problem, only good results.


> 
>> A port that does not support staging by my birthday gets deprecated.
> 
> Agreed; but it seems people are stagifying them as a pretty fast pace
> already, so it is not really a problem.

I take issue with this.  If you were one that was doing all this extra
staging I'd be fine with this statement, but last I looked you had over
60 of your own ports that need stage support.  At this rate, somebody
will have to do that work for you, so I really take issue that you are
relying on this work by others for rate.  I'm guilty too but at least
all my own ports (~35) have been staged for a long time.  I'm still not
taking this volunteer effort for granted though.


>> Any such ports that have been deprecated for two months and not seen
>> any work to fix them get removed.
> 
> I still don't see the reason to remove ports so promptly.  I would say
> half year looks more feasible to me; it also gives more time to build
> clusters to recover from occasional sporadic, transient, or network
> errors.


Most of the ports getting culled have had WAY longer than 6 months to
get right.  Resurrection is the real way of knowing if anybody finds
value in the port.  Most will stay dead wthout anybody making a peep.
All the ports are saved by SVN, I'm not seeing any sort of problem.

John