Re: [PATCH] Use role name "system_user" instead of "user" for unsafe_tests - Mailing list pgsql-hackers

Hi,

While playing with a new single board computer (VisionFive 2) I
discovered that postgresql:unsafe_tests suite fails like this:

```
--- /home/user/projects/postgresql/src/test/modules/unsafe_tests/expected/rolenames.out
2023-04-11 14:58:57.844550612 +0000
+++ /home/user/projects/postgresql/build/testrun/unsafe_tests/regress/results/rolenames.out    2023-04-11 17:54:22.999024391 +0000
@@ -53,6 +53,7 @@ CREATE ROLE "current_user"; CREATE ROLE "session_user"; CREATE ROLE "user";
+ERROR:  role "user" already exists RESET client_min_messages; CREATE ROLE current_user; -- error ERROR:  CURRENT_USER cannot be used as a role name here
@@ -1089,4 +1090,5 @@ DROP OWNED BY regress_testrol0, "Public", "current_role",
"current_user", regress_testrol1, regress_testrol2, regress_testrolx
CASCADE; DROP ROLE regress_testrol0, regress_testrol1, regress_testrol2,
regress_testrolx; DROP ROLE "Public", "None", "current_role", "current_user",
"session_user", "user";
+ERROR:  current user cannot be dropped DROP ROLE regress_role_haspriv, regress_role_nopriv;
```

This happens because the developers of this SBC choose the default
username "user", which I had no reason to change.

Test merely checks that we can distinguish a username "user" from the
USER keyword. Maybe it's worth replacing "user" with "system_user"? It
is also a keyword but is a less likely choice for the OS user name.

I don't think we can protect against all possible user names. Wouldn't it be better to run the tests under an OS user with a different name, like "marmaduke"? ("user" is a truly terrible default user name).

cheers

andrew

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