cfbot vs. changes in the set of CI tasks

On Fri, Sep 23, 2022 at 9:44 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:

I was confused about how come the new patches I'd just posted in
the 3848 CF item (write/read support for raw parse trees) are
showing a mix of passes and fails in the cfbot. I eventually
realized that the fail results are all old and stale, because
(for example) there's no longer a "FreeBSD - 13" task to run,
just "FreeBSD - 13 - Meson". This seems tremendously confusing.
Can we get the cfbot to not display no-longer-applicable results?

As a quick-fix hack it might do to just flush all the pre-meson-commit
results, but I doubt this'll be the last time we make such changes.
I think an actual filter comparing the result's time to the time of the
allegedly-being-tested patch would be prudent.

Ah, right. Morning here and I've just spotted that too after the
first results of the Meson era rolled in overnight. It shows the
latest result for each distinct task name, which now includes extinct
tasks (until they eventually get garbage collected due to age in a few
days). I clearly didn't anticipate tasks going away. Perhaps I
should figure out how to show only results corresponding to a single
commit ID... looking into that...

On Fri, Sep 23, 2022 at 10:17 AM Thomas Munro <thomas.munro@gmail.com> wrote:

On Fri, Sep 23, 2022 at 9:44 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:

I was confused about how come the new patches I'd just posted in
the 3848 CF item (write/read support for raw parse trees) are
showing a mix of passes and fails in the cfbot. I eventually
realized that the fail results are all old and stale, because
(for example) there's no longer a "FreeBSD - 13" task to run,
just "FreeBSD - 13 - Meson". This seems tremendously confusing.
Can we get the cfbot to not display no-longer-applicable results?

As a quick-fix hack it might do to just flush all the pre-meson-commit
results, but I doubt this'll be the last time we make such changes.
I think an actual filter comparing the result's time to the time of the
allegedly-being-tested patch would be prudent.

Ah, right. Morning here and I've just spotted that too after the
first results of the Meson era rolled in overnight. It shows the
latest result for each distinct task name, which now includes extinct
tasks (until they eventually get garbage collected due to age in a few
days). I clearly didn't anticipate tasks going away. Perhaps I
should figure out how to show only results corresponding to a single
commit ID... looking into that...

Done, and looking more sane now.

Thomas Munro <thomas.munro@gmail.com> writes:

On Fri, Sep 23, 2022 at 10:17 AM Thomas Munro <thomas.munro@gmail.com> wrote:

Ah, right. Morning here and I've just spotted that too after the
first results of the Meson era rolled in overnight. It shows the
latest result for each distinct task name, which now includes extinct
tasks (until they eventually get garbage collected due to age in a few
days). I clearly didn't anticipate tasks going away. Perhaps I
should figure out how to show only results corresponding to a single
commit ID... looking into that...

Done, and looking more sane now.

Looks better to me too. Thanks!

regards, tom lane

Hi,

On 2022-09-22 17:44:56 -0400, Tom Lane wrote:

I was confused about how come the new patches I'd just posted in
the 3848 CF item (write/read support for raw parse trees) are
showing a mix of passes and fails in the cfbot. I eventually
realized that the fail results are all old and stale, because
(for example) there's no longer a "FreeBSD - 13" task to run,
just "FreeBSD - 13 - Meson". This seems tremendously confusing.
Can we get the cfbot to not display no-longer-applicable results?

Somewhat tangentially - it seemed the right thing at the time [TM], but now I
wonder if tacking on the buildsystem like I did is the best way? As long as we
have both I think we need it in some form...

Greetings,

Andres Freund

Andres Freund <andres@anarazel.de> writes:

On 2022-09-22 17:44:56 -0400, Tom Lane wrote:

I was confused about how come the new patches I'd just posted in
the 3848 CF item (write/read support for raw parse trees) are
showing a mix of passes and fails in the cfbot. I eventually
realized that the fail results are all old and stale, because
(for example) there's no longer a "FreeBSD - 13" task to run,
just "FreeBSD - 13 - Meson". This seems tremendously confusing.
Can we get the cfbot to not display no-longer-applicable results?

Somewhat tangentially - it seemed the right thing at the time [TM], but now I
wonder if tacking on the buildsystem like I did is the best way? As long as we
have both I think we need it in some form...

Yeah, I think those names are fine for now.

regards, tom lane