3,058 reputation
1429
bio website None
location United States
age 53
visits member for 4 years, 11 months
seen Jul 2 at 16:19
Long-time Informix user and developer, experienced in C and Unix (many variants). Email: jonathan.leffler@gmail.com

Aug
19
awarded  Popular Question
Jul
26
comment Why can't I mount the same filesystem at multiple points, and why can't a mount-point inode reference count be > 1?
Two of about seven — an exaggerated version of many. I prefer that the addendum is a comment (or a separate answer) than a modification of my answer.
Jul
26
revised Why can't I mount the same filesystem at multiple points, and why can't a mount-point inode reference count be > 1?
Remove the as yet unjustified addition to my answer.
Jul
26
comment Why can't I mount the same filesystem at multiple points, and why can't a mount-point inode reference count be > 1?
@Neil Smithline: Please justify your 'many' variants by citing at least one non-Linux variant that provides what you describe: '25 July 2012 Update: Many Unix/Linux variants now support the -o union option that allows two filesystems to share a mount point. The second mount shadows the first in that all file accesses take place first in the new mount and then in the original mount. Also, file creation happens in the new mount.' Wikipedia has a brief outline but little specificity about which o/s other than Plan 9 support them.
Jul
21
awarded  Yearling
Jul
20
awarded  Constituent
Jul
9
awarded  Caucus
Jun
8
awarded  Constituent
Jun
8
awarded  Caucus
May
30
awarded  Nice Answer
May
29
comment What is the reason for rmdir(1) and rm(1) to co-exist?
One major downside of the mknod + link and unlink system was that creating a directory was not an atomic operation, so you could end up with a partially complete directory. There were lots of programs devised to check file systems for the inconsistencies that arose; fsck(1) is the one that survived.
May
29
comment What is the reason for rmdir(1) and rm(1) to co-exist?
Actually, back in the early days of Unix, neither rmdir(2) nor mkdir(2) existed as a system call; user root could use the mknod(2) call to create a directory node and the link(2) call to create the . and .. entries in the directory; and root could use the unlink(2) call to remove the directory entries.
May
29
answered What is the reason for rmdir(1) and rm(1) to co-exist?
May
14
awarded  Nice Answer
May
3
comment Does the target in ln -s have to be a full path?
You ask about ln -s but your example uses just ln; which are you dealing with?
May
1
comment Unix tools: what if a file is named minus something?
To the 'close because off-topic' voter: This is a question about shell programming and how to avoid a problem. It is totally on-topic for SO. (OTOH, it is probably a duplicate; the problem is finding the relevant other question.)
Apr
29
comment How to get infinite command history in bash?
You could probably cut that down to 9999 in practice (or even 999; I use 1000). I've never found a history bigger than about 1000 to be useful. If I want to recover a command to make it into a script, I normally do that well within a thousand commands. Otherwise, it can go and there's no harm done.
Apr
24
answered Installing a second version of e.g. git, perl and or python on osx lion
Apr
6
awarded  Popular Question
Jan
31
answered Sudo segmentation fault in Terminal