On Tue, 2025-01-07 at 07:40 -0800, Jeff Johnson wrote: > On 1/7/2025 3:12 AM, Johannes Berg wrote: > > On Tue, 2025-01-07 at 16:03 +0530, Sarika Sharma wrote: > > > Actually, there is ongoing work to extend sinfo for link-level details. > > > While working on this, I noticed a documentation mismatch and decided to > > > correct the existing documentation first, as it's good to have aligned > > > structure and documentation. > > > > Why do you think it's good? I don't even agree with that. The code > > should be laid out to minimize holes, but the docs can be laid out to > > group functionally related fields. > > Without any documented guidance to that effect Yeah I don't see anything either? Though kernel-doc certainly could complain about the order if it wanted to. I tend to think it doesn't need to match though, per above. > I had suggested this during > internal review to help minimize the size of the diffs where members are being > refactored out of sinfo into a new per-link struct. Fair enough. > But we can drop this since I think we'll be renaming some of the refactored > members and hence the diff will be big anyway. I think we can also apply it if you think it makes things easier, I just didn't think "it should match" really is a good justification. I don't care too much about the order in the docs, but also don't see needlessly changing it as a useful thing. johannes
