Re: CVS commit: pkgsrc/databases

[Aleksej Saushev <asau%inbox.ru@localhost>]

"OBATA Akio" <obache%netbsd.org@localhost> writes:

> 1. remove completely PG_SUBPREFIX support, then add CONFLICTS

We do not need any CONFLICTS in packages that either
a) have support staged installation or
b) have static PLIST.
There's no technical justification to keep it.
Package conflict is a fact that is detected automatically already.

> or
> 2. revive improved PG_SUBPREFIX support, then avoid CONFLICTS
>
> On Tue, 23 Oct 2012 06:50:48 +0900, Aleksey Cheusov 
> <cheusov%tut.by@localhost> wrote:
>
>>> I don't know it is it is expected, but PG_SUBPREFIX will only affect to
>>> postgresql90.
>> You are right. PG_SUBPREFIX is also mentioned in other PGs but
>> set only in PG90.
>>
>> Have a look to what was in the previous revision.
>>
>> http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/databases/postgresql90/Makefile.common.diff?r1=1.20&r2=1.21&only_with_tag=MAIN&f=h
>>
>> postgresql-90-* (including pg90 modules) conflicted with everything,
>> i.e. pg83, pg84, pg91 and pg92 (also including all possible modules).
>> PG_SUBPREFIX variable was completely ignored.
>> So, my patch doesn't introduce any regression.
>>
>> As about PG_SUBPREFIX, I think in its current form it should just be
>> completely removed and reimplemented in a different way, if we want to
>> avoid conflicts between different PGs. For example, install *all* PGs
>> with uniq subprefix by default and implement a meta package
>> databases/postgresql with pkg_alternative mechanism for mapping
>> executables to bin/.
>>
>
>
> --
> OBATA Akio / obache%NetBSD.org@localhost
>

-- 
HE CE3OH...