Hello Jakub, > > Provide some suggestions that who deal with Netlink code could follow > > (of course using the word "best-practices" may sound somewhat > > exaggerate). > > I truly appreciate the effort, but I'm worried this will only confuse > developers. This document does not reflect the reality of writing new > netlink families at all. It's more of a historic perspective. While I understand your concerns, I have a slightly different perspective on the statement that it "does not reflect the reality... at all." The "About General Netlink Case" section does highlight some important considerations for writing new netlink families, such as using helpers to access nlattr and avoiding deprecated parsers. However, I do acknowledge that the second best practice could be confusing, as developers who use auto C code generation may not encounter the issues mentioned. Moving forward, I suggest we consider the following options: 1. Update the document to address the confusion and make it more relevant to the current state of Netlink development. Maybe the newly added section seems not enough for that. I would greatly appreciate any specific guidance. 2. If the document is deemed too outdated for being kernel documentation, maybe I should publish it somewhere else. Do you have any recommendations on where it could be better suited? Thank you for your time and consideration. Have a great weekend. :D Regards Lin
