On 18/12/2025 09:21, Maximiliano Sandoval wrote:
> Yes, but the quote is from the nm-settings man page. GLib's .ini format
> allows to decide which separator to use **globally**, see [2], it does
> not allow for some lists will be handled with one separator while others
> with another, not without out-of-tree parsing at least. There are some
> exceptions in nm-settings though, but they are mentioned in the man page
> and are not handled as lists, namely:
>
> ```
> Also, some lists of complex values (addresses, routes, routing-rules),
> instead of using a semicolon separated list, use one key-value pair per
> list element,
> ```
>
> [2] https://docs.gtk.org/glib/method.KeyFile.set_list_separator.html
Oof... But in this should IMO then get added to the commit message, as
when I check the docs and see examples with either variant, but without
having mentioned that it depends which variant is OK when, then any
rationale in the commit messages helps me a ton to figure that out for
myself quickly.
Do you mind re-sending this with the commit message amended? Maybe we can
also introduce a local glib_join_list helper that is basically just a
wrapper like sub { return join(';', @_); } with a comment above?
We have quite some distros we support here and things do change but in
varying rates, so such semantic helpers can be quite nice to have.
Adding already a test case would be also appreciated, hopefully the
"src/test/test-centos10-001" one can be copied and adapted to cover
this case here.
_______________________________________________
pve-devel mailing list
[email protected]
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
