On Fri, Apr 22, 2022 at 10:24 AM Andrew Dunstan <and...@dunslane.net> wrote: > On 2022-04-22 Fr 10:04, Tom Lane wrote: > > Andrew Dunstan <and...@dunslane.net> writes: > >> This has upset the buildfarm's msys2 animals. There appears to be some > >> wildcard expansion going on that causes the problem. I don't know why it > >> should here when it's not causing trouble elsewhere. I have tried > >> changing the way the tests are quoted, without success. Likewise, > >> setting SHELLOPTS=noglob didn't work. > >> At this stage I'm fresh out of ideas to fix it. It's also quite possible > >> that my diagnosis is wrong. > > When I was looking at this patch, I thought the number of test cases > > was very substantially out of line anyway. I suggest that rather > > than investing a bunch of brain cells trying to work around this, > > we just remove the failing test cases. > > WFM.
Sure, see also http://postgr.es/m/CA+TgmoYRGUcFBy6VgN0+Pn4f6Wv=2h0hzluphqsy6vc8ba7...@mail.gmail.com where Andrew's opinion on how to fix this was sought. I have to say the fact that IPC::Run does shell-glob expansion of its arguments on some machines and not others seems ludicrous to me. This patch may be overtested, but such a radical behavior difference is completely nuts. How is anyone supposed to write reliable tests for any feature in the face of such wildly inconsistent behavior? -- Robert Haas EDB: http://www.enterprisedb.com
