as expected, a missing PSEUDO_LOCALSTATEDIR throws fatal error $ PSEUDO_LOCALSTATEDIR=/a/b/c/d ./result/bin/pseudo Warning: PSEUDO_PREFIX unset, defaulting to /tmp/./result. Can't open local state path '/a/b/c/d': No such file or directory pseudo-1.9.0/pseudo_client.c pseudo_diag("Can't open local state path '%s': %s\n", but if PSEUDO_LOCALSTATEDIR is read-only then pseudo produces this deadloop $ ./result/bin/pseudo Warning: PSEUDO_PREFIX unset, defaulting to /tmp/./result. help: can't open log file /tmp/./result/var/pseudo/pseudo.log: Read-only file system help: can't open log file /tmp/./result/var/pseudo/pseudo.log: Read-only file system help: can't open log file /tmp/./result/var/pseudo/pseudo.log: Read-only file system ... pseudo-1.9.0/pseudo_util.c pseudo_diag("help: can't open log file %s: %s\n", pseudo_path, strerror(errno)); i tried adding "exit(1);" after the pseudo_diag call, but no help so this condition should be checked earlier in the code
Is this an edge case bug for pseudo or is there an oe-core workflow that breaks due to this problem?
its just a bug in pseudo i have no idea what oe-core workflow is
Ah right. I forgot that some people use pseudo outside of Yocto builds. Can you send a patch to: ❯ rg sendemail .../pseudo.git/README 97: git config sendemail.to yocto-patches@lists.yoctoproject.org ?
i have no time for patching, sorry
Deadlocks are bad so we should fix this at some point but read only filesystems are not common so medium priority.
Bulk move of all unassigned 5.2 medium importance bugs to 5.3.
I don't think this has been fixed. Bumping to 6.1.