Conversation
This way subtests can stand alone rather than depending on each other. Also, fix several expectations. Comments indicate that the correct behavior was not observed, but rather than annotate it with a "todo" statement the assertions were altered to expect incorrect behavior instead. But in fact the correct behavior is observed when starting with a fresh ACL for each subtest.
ngie-eign
left a comment
ngie-eign
left a comment
There was a problem hiding this comment.
/bin should be in $PATH.
Does this test work on Linux or should it be made to work on Linux/ZFS?
|
Sorry, but it's been too long since I worked on this. And I think that nowadays my time would be spent more productively on other projects. Would you like to take over this PR, @ngie-eign ? |
Sure -- I can do that, but I need write access to the repo. |
I just invited you to collaborate on my fork. |
|
The test doesn't even seem to pass as-is on FreeBSD 15.1-RELEASE prior to your changes :(... |
|
I've been walking through the tests over the past few hours and I'm trying to isolate each of them into their respective "assertion groups" to help eliminate cascading failures present in 15.1-RELEASE, which your proposed changes don't address since they're issues with ZFS as-is in FreeBSD. |
|
I think it's worth just taking the tests, distilling them down to their expectations (as shorthand testplans), then convert it over to the rust-equivalent tests since the tests don't function as-is against FreeBSD 15.1-RELEASE with the default settings noted in zfsprops(7). I'll look at doing that with #88 . |
2 participants
This way subtests can stand alone rather than depending on each other.
Also, fix several expectations. Comments indicate that the correct
behavior was not observed, but rather than annotate it with a "todo"
statement the assertions were altered to expect incorrect behavior
instead. But in fact the correct behavior is observed when starting
with a fresh ACL for each subtest.