1 |
On 03/21/2018 03:33 AM, Kent Fredric wrote: |
2 |
> On Wed, 21 Mar 2018 01:44:11 +0000 |
3 |
> Herb Miller Jr. <herb@×××××.com> wrote: |
4 |
> |
5 |
>> If I am, then yes, some kind of automation |
6 |
>> would be the only sane way to keep up |
7 |
> In my experience you can't *really* rely on automation 100% for this |
8 |
> sort of thing. Not while achieving quality results. |
9 |
> |
10 |
> Its viable for an overlay where there's no expectations of quality, but |
11 |
> for the main tree, I find you want to have a human san-check everything |
12 |
> and manually vet each upstream version for "anomalous things". |
13 |
> |
14 |
> Automation is good at handling the "known predictable" cases, humans |
15 |
> are better at detecting "huh, that's weird, why did they do that?" |
16 |
> |
17 |
> Because you absolutely want to know if upstream added some stupid |
18 |
> change that is harmful to Gentoo users before you blindly replicate it. |
19 |
|
20 |
And I agree with you 100%. I would never rely on automation exclusively. |
21 |
I would use it to write boilerplate sections, check for updates, check |
22 |
for breakage, etc. I'd never open a PR for something I hadn't polished |
23 |
myself. |