| 1 |
On 12/07/17 03:16, William Hubbs wrote: |
| 2 |
> On Tue, Jul 11, 2017 at 11:47:32PM +1000, Michael Palimaka wrote: |
| 3 |
>> On 07/11/2017 11:06 PM, Rich Freeman wrote: |
| 4 |
>>> On Tue, Jul 11, 2017 at 8:59 AM, Michael Palimaka <kensington@g.o> wrote: |
| 5 |
>>>> On 07/11/2017 09:29 AM, Andrew Savchenko wrote: |
| 6 |
>>>>> |
| 7 |
>>>>> Even if such stabilization is allowed, there are unanswered |
| 8 |
>>>>> questions here: |
| 9 |
>>>>> - is following seciton 4.1 from wg recommendations is sufficient? |
| 10 |
>>>>> - should developer test each stabilization candidate on an |
| 11 |
>>>>> up-to-date stable setup? |
| 12 |
>>>> |
| 13 |
>>>> The guidelines from that document are ripped straight out of the |
| 14 |
>>>> devmanual and are a good starting point but rather generic. You can find |
| 15 |
>>>> some more detailed suggestions on things to consider while testing on |
| 16 |
>>>> the wiki: https://wiki.gentoo.org/wiki/Package_testing |
| 17 |
>>>> |
| 18 |
>>> |
| 19 |
>>> I think that in practice arch teams don't do have the stuff on that |
| 20 |
>>> wiki page. Maybe some people do, but back when I was an amd64 AT I |
| 21 |
>>> don't think anybody went testing multiple USE combinations for a |
| 22 |
>>> typical package. |
| 23 |
>> |
| 24 |
>> Everything on that page is deliberately a suggestion only, and not |
| 25 |
>> necessarily specific to stabilisation testing. |
| 26 |
>> |
| 27 |
>> In the end, we've never been able to reach any consensus on what exactly |
| 28 |
>> an arch tester should do. Personally, I think we should just switch to |
| 29 |
>> fully-automated, build-only testing for stabilistions unless the |
| 30 |
>> maintainer opts otherwise (something that largely happens in practice |
| 31 |
>> already). The main risk of breakage of a package moving from testing to |
| 32 |
>> stable is always at build time anyway. |
| 33 |
> |
| 34 |
> I would not be opposed to this. As a maintainer, I am as guilty as the |
| 35 |
> next guy of not filing stable requests or not stabilizing packages. |
| 36 |
|
| 37 |
As the next guy, I also +1 this. |
| 38 |
|
| 39 |
As is often explained in #gentoo, "stable" and "testing" for upstream |
| 40 |
have a different meaning to "stable" and "testing" for Gentoo - in fact, |
| 41 |
beyond ensuring it builds, any testing performed by a downstream distro |
| 42 |
is additional testing to what upstream have already released. |
| 43 |
|
| 44 |
I can understand the concern with automatically stabilising a package |
| 45 |
that has some flaw affecting users, but the two things i see are that |
| 46 |
upstream have already released said flawed package, and Gentoo simply |
| 47 |
doesn't have the manpower to perform comprehensive runtime testing of |
| 48 |
all packages. |
| 49 |
|
| 50 |
If a maintainer is aware of a significant issue with a package that |
| 51 |
should prevent it from being marked as stable, then a bug should be |
| 52 |
filed acting as a block to the automated stabilisation. Users aware of |
| 53 |
an issue are just as likely to file a bug as well, also preventing |
| 54 |
stabilisation of packages with some known issue. Therefore packages with |
| 55 |
known issues don't get stabilised automatically. |
| 56 |
|
| 57 |
Similarly, if the maintainer believes more comprehensive testing should |
| 58 |
be done (eg. for critical base-system packages) a note could be made |
| 59 |
somewhere of that requirement (metadata.xml?), meaning significantly |
| 60 |
reduced workload for people like ago (if the maintainer doesn't |
| 61 |
stabilise it beforehand). |
| 62 |
-- |
| 63 |
Sam Jorna (wraeth) |
| 64 |
GnuPG ID: D6180C26 |
Archives