[oi-dev] Testing Updating packages and hipster-testing repo.

Nikola M. minikola at gmail.com
Wed Oct 9 06:55:17 UTC 2013


On 10/ 9/13 08:37 AM, Alexander Pyhalov wrote:
> On 10/09/2013 10:28, Adam Števko wrote:
>> Hi,
>>
>> are there any problematic packages, which don't work with never PHP? 
>> If no, go for it.
>>
>
> I don't know about problematic packages. However, it doesn't mean that 
> they don't exist.
>
> I'm just looking at Ken request about updating net-snmp and going 
> through net-snmp dependencies.
I think that every change to public Hipster repo should be tested before 
done.
And it goes for every update or group of them, so it could be tested by 
one more or few people, before pushing it public.

How about pushing it to hipster-testing publisher or something, before 
going to hipster?
If Hipster is rolling-release, it should have 'release' in mind, not 
only 'rolling' :)

First contributor should check dependent packages, then verify on it's 
own install and local publisher how it works,
then push to hipster-testing (possibly with instructions and some 
information and observation how he think it should be tested and what) 
and after some period of time and confirmation that it is tested,
it could go to hipster publisher.  (testing could include checking if 
update from not only previous version, but also how update whole Hipster 
from what OI /dev or Hipster point release is going.)

That way it could be avoided for stupid bugs to surface to the people 
(meaning it did not work in the first place before pushing to the 
Hipster and no one checked) and see that important bugs surface on 
actual use of updated package and not packaging/dependencies problems etc.

What do you think?

Nikola M.





More information about the oi-dev mailing list