On Thu, 12 Dec 2024 11:41:26 GMT, Hannes Wallnöfer <hann...@openjdk.org> wrote:
> > > Omit type bounds and do not create separate links for type parameters in > > > links to inherited nested classes > > > > > > But I also wonder if the below is acceptable information loss. > > In my opinion, it is not just acceptable but a clear improvement. Consider > the benefits of the additional links for type parameters and bounds: > > * The type parameter links are almost completely redundant with the link to > the encosing type, as type parameters are documented prominently at the top > of the type documentation. > * The bound on the last type parameter in this particular case doubles all > the links again, leading to 8 links to basically the same resource (except > for highlighting individual type parameters). > > I admit that the recursive bound is a special case, and a link would make > more sense if it was a different type. But even in that case I would argue > that it's not at the link level that we have to provide these details, and in > fact we don't do that in other similar places (e.g. lists of > super/implementing/extending types etc). > > As to the cost of providing these additional links in this context: there is > a cost with every link we add in terms of demanding attention for the user, > but in this case it's particularly bad, because the context (list of > inherited nested classes) and the content (linked signature of generic class) > both basically use the same format (a comma-separated list of links). Which > is how we end up with a hairball of largely redundant links when we really > wanted to provide a link to one nested class. Alternatively, we can indicate that the class or interface declaration has more interesting to it by maybe placing … (`…`). ------------- PR Comment: https://git.openjdk.org/jdk/pull/22651#issuecomment-2538713103
