On Thu, Aug 25, 2022 at 05:54:25PM -0500, G. Branden Robinson wrote: > [sorry for the big CC] > > At 2022-08-25T11:06:55-0700, Alexei Starovoitov wrote: > > Nack. > > > > We don't follow 80 char limit and are not going to because of man > > pages. > > If someone got a contract with O'Reilly or No Starch Press to write a > book on BPF and how revolutionarily awesome it is, it's conceivable they > would be faced with exposing some BPF-related function declarations in > the text. In cases like the following, what would you have them do? > > int bpf_map_update_elem(struct bpf_map *map, const void *key, const void *value, u64 flags) > > Be aware that the author may not have infinite flexibility; publishers > generally impose a "house style" which restricts selection of typeface > (so they can't necessarily print at a smaller type size or with the > kerning reduced beyond a certain point to squeeze the text onto the > line). As someone who has written a book for one of those publishers you mention above, this is totally not an issue at all. Authors and typesetters know how to properly wrap and handle stuff like this automatically, it's what they do and has nothing to do with how kernel header files are layed out. But even then, if it was an issue, we don't write kernel code for "some potential commercial entity that can't handle long lines", you know better than this :) greg k-h