Le ven. 29 mai 2020 à 15:20, Arthur Miller a écrit : > Richard Stallman writes: > > > [[[ To any NSA and FBI agents reading my email: please consider ]]] > > [[[ whether defending the US Constitution against all enemies, ]]] > > [[[ foreign or domestic, requires you to follow Snowden's example. ]]] > > > > > This seems like a reasonable solution to me. Alternatively perhaps we > > > just need to sell C-x C-f as "open a file or directory" rather than > > > "find a file"? > > > > That would make our initial explanations more complex, and that might > > lead to more confusion than clarity. > > > > I think it is better to explain this wrinkle when the user encounters it, > > not before. > > Aren't users encountering that wrinkle first time they open a file? > > Observe there are even more wrinkles there to explain: if file does not > exist Emacs creates a new buffer, and if user ment a directory, the > buffer will still be just a plain file not a dir. And what about if > there are some non-existent directories on the way? Emacs asks if user > wants them to be created ... so there are quite a few wrinkles in that > one, not the simplest behaviour to explain anyway :-). > But those are consistent with how other software behaves. The user won't think of a buffer as much as of a not-yet-saved file, but that's also consistent with what emacs does. Other software would force the directory to be created before opening the file, but that's minor I believe. Being able to open a directory just like a file, on the other hand, is not usual. Web browsers can do it, but that's because their "files" are almost like directories. I personally like the ido approach of having different keys for accessing files and directories. Simple operations like listing files or creating a new one are done directly in the ido buffer, no need for dired. Thibaut