svn commit: r516880 - in head: archivers/xarchiver astro/garmindev astro/opencpn astro/qmapshack audio/carla audio/decibel-audio-player audio/festvox-czech audio/gkrellmvolume2 audio/gnaural audio/...
Craig Leres
leres at freebsd.org
Sun Nov 10 00:45:10 UTC 2019
On 2019-11-09 02:19, Mathieu Arnold wrote:
> On Fri, Nov 08, 2019 at 04:20:59PM -0800, Craig Leres wrote:
>> On 2019-11-08 14:02, Antoine Brodin wrote:
>>> On Fri, Nov 8, 2019 at 8:40 PM Craig Leres <leres at freebsd.org> wrote:
>>>>
>>>> On 2019-11-06 04:48, Antoine Brodin wrote:
>>>>> Author: antoine
>>>>> Date: Wed Nov 6 12:48:32 2019
>>>>> New Revision: 516880
>>>>> URL: https://svnweb.freebsd.org/changeset/ports/516880
>>>>>
>>>>> Log:
>>>>> Mark a few ports BROKEN, unfetchable
>>>>>
>>>>> Modified:
>>>>
>>>>> head/devel/xtensa-esp32-elf/Makefile
>>>>
>>>> I just did:
>>>>
>>>> cd devel/xtensa-esp32-elf
>>>> rm distinfo
>>>> make makesum
>>>>
>>>> and was able to download all of the DISTFILES again and generate a new
>>>> distinfo that is identical except for the TIMESTAMP line.
>>>>
>>>> What identified this port as unfetchable?
>>>
>>> You must be joking.
>>
>> (That's not very friendly or helpful!)
>
> Well, the thing is, portmgr's job is not to fix all the ports, it is to
> curate the ports tree. It means that when something is broken, we don't
> try to look at why. We mark it broken and let the community fix it.
>
> Also, we only mark ports as being broken because they are not fetchable
> after about six months of them not being fetchable.
I guess my point is if port maintainers could be notified when a port
becomes unfetchable it would give them time to fix the problem. (Perhaps
six months?) Consider that users of the port will have a local copy of
the DISTFILES that work in /usr/ports/distfiles. So if something happens
with individual MASTER_SITES we're not going to notice.
> And each time, people who should know better assume that we're dumb and
> are in error because while they should know how the ports tree work,
> they figure they know better than us and we're wrong.
>
> So, yeah, from time to time, especially after it's the fifth time in
> like two days that someone tells us "you're doing shit" while being
> clueless, it happens that we're not as patient as we could be.
I'm 200% sure I didn't tell anybody "you're doing shit". I asked what
how the port was selected and this was (eventually) explained.
But who runs this process? How often does it happen? For example you
marked net-mgmt/telegraf as BROKEN/unfetchable. But I just ran "make
checksum DISTDIR=/tmp" and was able to re-download all 146 files in
distinfo. So while I'd like to port to be buildable again I'm not sure I
understand what is wrong with it. Was a temporary server problem? Is it
fetchable/not-fetchable depending on where you try to download it? Was
it just a mistake?
Meanwhile I'll ask again: is possible to get advance warning about
unfetchable probelms? It sounds like portscout should do it or used to
do it. How do you generate the list of BROKEN/unfetchable? Is it just
adhoc or is there a process that systematically identifies candidates?
Please understand I'm not complaining, just trying to understand how I
can do a better job and avoid having ports that are important to me
broken at inconvenient times.
Craig
More information about the svn-ports-head
mailing list