Re: PostgreSQL PKGNAME

[David Brownlee <abs%netbsd.org@localhost>]

On 17 December 2010 21:14, Greg Troxel <gdt%ir.bbn.com@localhost> wrote:
>
> Joerg Sonnenberger <joerg%britannica.bec.de@localhost> writes:
>
>> On Fri, Dec 17, 2010 at 10:41:03AM -0500, Greg Troxel wrote:
>>>
>>> Do you object to the act of changing, or to the new state? ÂCould you
>>> explain the 'major hassle'?
>>
>> Both, but primarily the new state. It basically requires any pattern an
>> explicit limit if you don't want to get surprised when a new Postgres
>> version arrives. In the real world, a Postgres update requires at least
>> some planing to deal with the database format change, if nothing else.
>
> OK, that's a good reason. ÂIt seems important for pgsql major version
> jumps to only happen when the box admin explicitly tries, and this seems
> more important than getting to avoid cluttered names and not having to
> have explicit CONFLICTS. ÂSo I will concur in your objection to removing
> the version tag from PKGNAME.

Maybe we could add a short comment to the relevant package Makefiles
documenting this reason - both to help anyone else raising this question
and to explain what kind of prior art they comprise if anyone is tempted
to copy the layout in future for other packages...