MinGW CI tasks fail / timeout

Hi,

On 2026-03-26 17:40:32 +0300, Nazir Bilal Yavuz wrote:

I noticed that MinGW CI tasks are currently failing and timing out
[1]. It appears that this started after the IPC::Run package was
updated on the MinGW CI image from: 'NJM/IPC-Run-20250809.0.tar.gz' to
'TODDR/IPC-Run-20260322.0.tar.gz' [2]. If I explicitly install
NJM/IPC-Run-20250809.0.tar.gz, then MinGW CI passes again [3].

I guess this PR [4] is responsible for the failures but I am not sure
about the fix so I wanted to bring that topic to -hackers.

FWIW, to mitigate the impact I've for now disabled running the mingw tests on
the cfbot and postgres repos. That's obviously not a real fix, but it seemed
better than spuriously failing all CF entries.

Feels somewhat crazy to just make a compatibility breaking change like that in
IPC::Run?

Greetings,

Andres Freund

On Thu, Mar 26, 2026 at 7:46 AM Andres Freund <andres@anarazel.de> wrote:

Feels somewhat crazy to just make a compatibility breaking change like that in
IPC::Run?

Looks like as of two weeks ago, IPC::Run is under somewhat new management?

--Jacob

On 2026-03-30 Mo 5:00 PM, Jacob Champion wrote:

On Thu, Mar 26, 2026 at 7:46 AM Andres Freund <andres@anarazel.de> wrote:

Feels somewhat crazy to just make a compatibility breaking change like that in
IPC::Run?

Looks like as of two weeks ago, IPC::Run is under somewhat new management?

Looks to me more like the old manager getting active again.

cheers

andrew

--
Andrew Dunstan
EDB: https://www.enterprisedb.com

Hi,

On Thu, 26 Mar 2026 at 17:46, Andres Freund <andres@anarazel.de> wrote:

On 2026-03-26 17:40:32 +0300, Nazir Bilal Yavuz wrote:

I noticed that MinGW CI tasks are currently failing and timing out
[1]. It appears that this started after the IPC::Run package was
updated on the MinGW CI image from: 'NJM/IPC-Run-20250809.0.tar.gz' to
'TODDR/IPC-Run-20260322.0.tar.gz' [2]. If I explicitly install
NJM/IPC-Run-20250809.0.tar.gz, then MinGW CI passes again [3].

I guess this PR [4] is responsible for the failures but I am not sure
about the fix so I wanted to bring that topic to -hackers.

FWIW, to mitigate the impact I've for now disabled running the mingw tests on
the cfbot and postgres repos. That's obviously not a real fix, but it seemed
better than spuriously failing all CF entries.

The version of IPC::Run is pinned to 'NJM/IPC-Run-20250809.0' on the
MinGW image [1]https://github.com/anarazel/pg-vm-images/pull/139. I tested the new MinGW image and confirmed that CI
passes [2]https://cirrus-ci.com/build/4605203913113600. I think we can re-enable the MinGW CI task on the CFBot
and Postgres repos.

[1]: https://github.com/anarazel/pg-vm-images/pull/139
[2]: https://cirrus-ci.com/build/4605203913113600

--
Regards,
Nazir Bilal Yavuz
Microsoft

On Fri, Apr 3, 2026 at 8:54 AM Nazir Bilal Yavuz <byavuz81@gmail.com> wrote:

The version of IPC::Run is pinned to 'NJM/IPC-Run-20250809.0' on the
MinGW image [1]. I tested the new MinGW image and confirmed that CI
passes [2]. I think we can re-enable the MinGW CI task on the CFBot
and Postgres repos.

Thank you! I opened a draft PR to pin the other versions as well;
let's see what the CI thinks...

--Jacob