On March 5, 2024 4:07 pm, Max Carrara wrote: > These tests attempt to cover all syntax quirks that the format of > 'ceph.conf' currently allows. > > One known exception however is the handling of "quoted" and "unquoted" > strings, as Ceph's own parser appears to not actually differ between > them either. Curiously, if a "quoted string" isn't able to be > parsed by Ceph's implementation, it goes on to parse it as an > "unquoted string" anyway. [0] In both cases, the result is the same - > the string is parsed with quotes.
I don't think this is true - quoted strings do behave differently w.r.t. in-line comments and line continuations inside the quoted string (both break parsing on the ceph side ;)), and in general a missing closing quote seems to be fatal. > Each test case is first tested against the parser - if the resulting > hash matches the expected hash, it is consequently passed into the > writer. The writer's result is then parsed another time and compared > against the expected hash once more. I think it would make a lot of sense to also test our parser against `ceph-conf -c TEST_INPUT_FILE [--name X.Y] --show-config-value KEY` in particular for all the edge cases surrounding quoting and so on it might help find some more discrepancies, but also help keeping current with ceph changes.. > [0]: > https://git.proxmox.com/?p=ceph.git;a=blob;f=ceph/src/common/ConfUtils.cc;h=2f78fd02bf9e27467275752e6f3bca0c5e3946ce;hb=refs/heads/master#l189 > > Signed-off-by: Max Carrara <m.carr...@proxmox.com> _______________________________________________ pve-devel mailing list pve-devel@lists.proxmox.com https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
