On Thu, Mar 6, 2025 at 4:38 PM Robert Haas <robertmhaas@gmail.com> wrote:
>
> On Thu, Mar 6, 2025 at 4:27 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> > My point was that we can implement such a policy in a laissez-faire
> > way: if an older BF animal isn't causing us trouble then why mess
> > with it? Once we *do* recognize that it's causing us trouble,
> > we can apply the still-hypothetical policy and ask the owner to
> > turn it off for branches where it's out of support.
>
> Fair enough. This does have the disadvantage that people will still
> commit things that turn the buildfarm red and have to go into panic
> mode -- perhaps not even realizing which machines are running an older
> OS -- and then reactively do this. However, it still sounds like
> progress over where we are today, because it would (hopefully) remove
> the need for an argument about each individual case.
One thing I've been wishing for recently is access to the discussion
and lore around individual buildfarm animals in a consolidated place.
As a new committer, I haven't been part of all of these discussions
over the last N years and so if an animal goes red, I can look through
the buildfarm status history to see what kind of failures it has had
for the last ~6 months, but it doesn't tell me if that machine is
known for failing because of an outdated OS or some other
platform-specific or hardware-specific issue.
For some animals searching for "skink" in the email archives might be
enough, but I haven't found that to be true so far.
- Melanie
