PR Load Solutions

[Matthias Andree]

Jesse Smith wrote on 2010-07-11:

> I'm not saying I have a solution. Just that I'd like to help out as much
> as the veteran maintainers are willing.

There's also the burden of testing PRs, we "veteran committers" (I still  
have two mentors myself) need to make sure the port builds and installs  
cleanly on the supported FreeBSD releases.

For that purpose, we set up so-called "tinderboxes" which builds ports in  
a controlled environment. If port maintainers can do that and provide the  
logs on a web-/fileserver and include the Tinderbox log URL in their PR,  
that may help a bit.


> On a related note, what about trying to actively attract upstream
> maintainers to help out with the ports of their projects? I didn't even
> know until recently that two of my projects had been added to the Ports
> tree. Once I found out, I wanted to help keep those ports maintained and
> up to date. Maybe other up-stream developers could be recruited to
> babysit their ports?

This has some drawbacks, especially for smaller upstream projects, so this  
should be decided case by case:

- if I am doing most of the upstream work, there are fewer eyes to look at  
the FreeBSD port;

- upstream maintainers may in some cases be less familiar with FreeBSD,  
they may not even use it. One such example is sysutils/e2fsprogs, another  
security/openvpn; just from my collection.

- upstream maintainers may be very good at programming, project  
management, whatever; FreeBSD port maintainers always cannot be too alien  
to systems administration.

- it usually pays off if the maintainer is actively using FreeBSD and the  
port he is maintaining. This is often not the case, otherwise the upstream  
maintainer already is the port maintainer :)

If this is done in the wrong way, it will backfire and actually raise  
support burden because the load of getting the actual "porting" part  
(FreeBSD adjustments) done propagates to committers...
Sure there are cases when the upstream maintainer is the port maintainer  
(f.i. news/leafnode, mail/bogofilter*), but I'm not sure this could fly as  
a general concept.

Note this is a personal opinion, not necessarily consensus. I'm /not/  
posting on behalf of FreeBSD here.

Best regards
Matthias

-- 
Matthias 'mandree@' Andree