svn commit: r332017 - head/devel/subversion17

[olli hauer]

On 2013-12-08 16:09, Matthias Andree wrote:
> Am 07.12.2013 15:07, schrieb Tijl Coosemans:
>> On Mon, 02 Dec 2013 18:03:16 +0100 Mathieu Arnold wrote:
>>> +--On 29 octobre 2013 22:01:05 +0000 Olli Hauer <ohauer at FreeBSD.org> wrote:
>>> | -PKGNAMESUFFIX=	17
>>> | +#7setting7PKGSUFFIX7not7possible7wo.7having7a7Mk/Uses7file7and7fixing7a7
>>> | bunch7of7ports
>>> | +#PKGNAMESUFFIX=	17
>>>
>>> I've grepped through the whole ports tree, and I couldn't find any port
>>> that depends on subversion1[67] using their package name, as in
>>> subversion<1.8:{$PORTS}..., they only depends on the svn binary being
>>> there, so adding PKGNAMESUFFIX will not hurt any bit.
>>>
>>> If I'm wrong, please show me which ports have their dependencies that needs
>>> fixing.
>>
>> The subversion-java slave port sets PKGNAMESUFFIX=-java, so setting
>> PKGNAMESUFFIX=17 in the master makefile would override that.
>>
>> The other slave ports (perl,ruby,python) set PKGNAMEPREFIX so maybe
>> the java port should do that too.
>>
> 
> Is there any purpose in keeping old garbage around?
> Such as new client not working with old server?
> If not, let's just axe the older versions rather than fix package name
> conflicts.
> 


The subversion java port is fixed, I'm more concerned with ports using

- ..._DEPENDS= subversion>=$version
- ..._DEPENDS= p5-subversion>=$version
- ..._DEPENDS= ${PYTHON_PKGNAMEPREFIX}subversion>=0:

Having PKGNAMEPREFIX will prevent building this ports in case default
subversion != devel/subversion even they work fine with subversion 1.6.

I'm going now over all ports depending on 'devel/subversion*' to check for
false declaration e.g using 'subversion>=1.7.0:' where 1.6 is sufficient.

Long time ago I've written a SVN framework, but got no or negative feedback
so I stopped the work. Hopefully I can find a bunch of time to start working
again on it.

http://lists.freebsd.org/pipermail/freebsd-ports/2011-July/068561.html