Date: Thu, 26 Apr 2018 14:25:47 +0200 From: Joerg Schilling <joerg.schill...@fokus.fraunhofer.de> Message-ID: <5ae1c54b.o3jqkvjsbowb+hep%joerg.schill...@fokus.fraunhofer.de>
| This is a miss interpretation. | The reason was just to avoid a mkdir() syscall. Jörg, that makes no sense at all. mkdir() was invented to deal with the atomicity issues with dealing with '.' and '..'. If those did not exist, mknod() would have been just fine for making directories (aside from it being root only, but that could have been changed for directories, if the '.' and '..' were not required - it was root only for directories so it they could only be made via the mkdir command). Please stop inventing history. | It is simple to write an interpreter in the kernel that understands the meaning | of the special files . and .. while doing path name resolution. Yes, it could have been done that way, provided one is willing to always make ".." mean parent, and not "wherever I have linked it to today". (I used to also make hard links to directories, which is why "parent made little sense, there was not always just one.) kre