On 2013-02-13 12:51:04, Michael P. Jung wrote:
> I'm starting to disbelieve myself. I wrote a simple C program that
> renames a file, creates a new one and forks a new process which tries to
> read the newly created file. It works just fine.
Well according to the strace, vim does remove the file for a moment:
That doesn't seem like the right thing to do to me, but that would be a bug in
vim.
I'm not familiar with how the auto reloader works, but there's a chance that it
could try to open "x" in the small amount of time that vim renamed "x" to "x~"
and hasn't yet recreated "x".
This may be only seen by you on eCryptfs because it is a little slower than
traditional filesytems.
On 2013-02-13 12:51:04, Michael P. Jung wrote:
> I'm starting to disbelieve myself. I wrote a simple C program that
> renames a file, creates a new one and forks a new process which tries to
> read the newly created file. It works just fine.
Well according to the strace, vim does remove the file for a moment:
rename("x", "/home/ bikeshedder/ .vimbackup/ x~") = 0 O_CREAT| O_TRUNC, 0664) = 3
fsync(4) = 0
open("x", O_WRONLY|
That doesn't seem like the right thing to do to me, but that would be a bug in
vim.
I'm not familiar with how the auto reloader works, but there's a chance that it
could try to open "x" in the small amount of time that vim renamed "x" to "x~"
and hasn't yet recreated "x".
This may be only seen by you on eCryptfs because it is a little slower than
traditional filesytems.