Created attachment 2691 [details] fix regress/hostkey_rotate.s During the walk through the regression test suite I hit few issues that should be fixed. Namely: 1) hostkey_rotation.sh contains some "mess" in the end of file, probably created as a skeleton for the test that was not used in the end (see patch) 2) keyscan.sh doesn't scan for ecdsa keys (nor for ed25519). All the other tests are now updated to reflect result of `ssh -Q key-plain`, or use the list imported from test-exec.sh, except this one. But sufficient would be to use just the list that is mentioned in manual page (“dsa”, “ecdsa”, “ed25519”, “rsa”).
The attached patch looks like it was made from the wrong files - it's for an Include option and doesn't contain any keyscan.sh bits. I've committed the hostkey-rotate.sh parts, thanks.
Yay. Sorry for mistake. I am working on more things at once again. But the first part in regress/hostkey-rotate.sh was understood well, the second part was not included in the patch, because I am not sure what is correct solution: Current version: > KEYTYPES="rsa dsa" > if ssh_version 1; then > KEYTYPES="${KEYTYPES} rsa1" > fi Documentation notes this list should work: > (“dsa”, “ecdsa”, “ed25519”, “rsa”) But it works well also for all the results from `ssh -Q key-plain`, as can be found in other test cases: > KEYTYPES=`ssh -Q key-plain` Anyway, the key-plain argument is missing in manual page for ssh so it would be great to add it.
Fixed (used ssh -Q key-plain and tweaked the manual) - thanks.
Close all resolved bugs after 7.3p1 release