Romain Francoise wrote:
For the record, why do you rebuild each snapshot? Not using unstable?
In the first weeks, that was one of the reasons as I was still running
sarge then; but not anymore. Now it has other reasons: following links in
*Help* buffers for built-in functions and variables (like
`buffer-backed-up' ;-)) requires the source code to be accessible, for
instance. I also build Emacs for other operating systems occasionally from
the same tarball. And since my Internet connection is not as fast as I
would wish, I also save some download time and costs by building from
source. The downside is that the build takes half an hour.
I will install your official binary emacs-snapshot packages and see if
the problem still appears. Anyway, it seems that I will have to debug
it myself if other people don't see it.
OK, please keep me posted. I'll tag this bug 'moreinfo' in the
meantime.
Now I have installed your binaries, am running
GNU Emacs 22.0.50.1 (i486-pc-linux-gnu, X toolkit, Xaw3d scroll bars)
of 2005-12-15 on pacem, modified by Debian
and can still see the problem. I have not figured out all the details yet,
but I can already give some additional information, namely the output of
"ls -li bar*" at different times:
Before running "emacs -Q -f view-hello-file bar" :
32625 -rw-r--r-- 1 sven sven 4 Dec 18 18:12 bar
32647 -rw-r--r-- 1 sven sven 4 Dec 18 18:12 bar~
After typing "C-x C-s" in Emacs, when the *Warning* pops up:
32625 -rw-r--r-- 1 sven sven 4 Dec 18 18:12 bar~
After quitting with "C-g":
32625 -rw-r--r-- 1 sven sven 4 Dec 18 18:12 bar
This explains clearly what I had seen: The saving attempt makes the backup
(by renaming, as is the default in Emacs), removing the old backup file in
the process and setting `buffer-backed-up' to t. And quitting renames the
file back, but does not touch `buffer-backed-up'.
Corollary: If `backup-by-copying' is t, only the old backup is lost, but
the new one remains.
I am really puzzled that this did not happen to you. The only circumstance
under which I did *not* see it was when I set `backup-inhibited' to t --
but you probably don't have such an insane setting, do you?
Can you please do "M-x debug-on-entry rename-file" and tell me if and when
this pops up the debugger in our scenario?
Regards,
Sven
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
