Archives of last quarterly package builds?

[Julian Elischer]

On 4/8/18 9:09 pm, Rainer Duffner wrote:
>
>
>> Am 04.08.2018 um 08:39 schrieb Kurt Jaeger <pi at freebsd.org 
>> <mailto:pi at freebsd.org>>:
>>
>> The idea is: use the quarterlies, and if the next quarter comes,
>> upgrade to that quarterly. The quarterlies are a way to test
>> if we can provide some 'more stable tree' than HEAD for the ports.
>>
>> It's not perfect, and we all learn the use cases and the issues etc.
>>
>> I don't have the overview over all the posts on that issues, so:
>> is there a text that describes alternative approaches ? Something
>> where implementation can be discussed ?
>
My issue is that just as it starts to get the bugs wrinkled out of it, 
it's deleted.

I gave up.  and now we mirror the ports tree in house and do a build 
of all the
packages at a given revision on the head branch,
and then OCCASIONALLY we slide a single package forward (or back) if 
we were
unfortunate in our snapshot and caught it with a bug/problem.
>
>
> The problem is that different people have different foci.
>
> I think it’s assumed that one hosts and maintains his (or her) own 
> copy of the ports-tree and maintains it according to one’s own 
> focus-points.
>
> E.g.: if I was to maintain my own fork of the ports-tree, I’d lay 
> the emphasis on a number of ports that greatly concern me (apache, 
> php, nginx, varnish, python and some of its base-ports, plugins for 
> nagios and some other stuff I’ve forgotten). I’d basically follow 
> upstream with those very closely.
> The rest, I’d let dormant most of the time, unless a 
> security-vulnerability made an update inevitable.
>
> But I’m really not in a position to do that, so I use the quarterly 
> cuts. They are a good compromise.
>
> Sometimes, I copy over a port from HEAD to my quarterly checkout 
> because I really want to have the update in. But that has become 
> rare, actually.
>
>
> Different people have different requirements.
> I think if you need very high stability, you’ll likely end up using 
> something else (CentOS+ Software Collections - or Ubuntu, if you’re 
> really desperate...)
>
> Certainly, someone from the foundation or some other company has 
> done the math on what it would take (man-power and financials) to 
> maintain certain subsets of the ports for longer than three months.
> Or everything.
>
> It will, however, be almost impossible to get it right for everybody.
>
>
>
>
>