On Mon, Dec 03, 2001 at 02:59:31PM -0500, David T-G wrote:
> ...and then Paul Brannan said...
> % This solves #2, but not #1 :(
>
> Interesting. What does
>
> mutt -v
>
> show you? Do you have dotlocking available? It looks to me as though
> mutt can only use fcntl locks but they are not supported on your system;
> mutt is usually pretty good at figuring out how it should lock.
The relevant portion of mutt -v indicates:
-HOMESPOOL +USE_SETGID +USE_DOTLOCK +USE_FCNTL -USE_FLOCK
I tried using mutt_dotlock on a test file in my home directory (as
suggested in the other thread), and it successfully created
testfile.lock; mutt_dotlock -u testfile successfully removed the file as
well.
I've never used fcntl before, but a simple test application yields:
fcntl: No locks available
The program looks like this:
int main() {
int fd = open("foo", O_CREAT | O_TRUNC | O_WRONLY);
struct flock l;
l.l_type = F_WRLCK; l.l_whence = 0; l.l_start = 0;
l.l_len = 0; l.l_pid = getpid();
if(fcntl(fd, F_SETLKW, &l) == -1) { perror("fcntl wr"); exit(1); }
return 0;
}
This same program works in /tmp (which is non-nfs; the first run the
program produces no output, and the second run the call to open()
fails).
Perhaps this is a problem of the client being able to lock over nfs but
the server not?
> If you want the messages to be deleted when you sync, set it to "yes" in
> your macro; if not, leave it at "no" and forget about the macro. If you
> want to be asked then just set it to ask-no and forget about the macro,
> and when you usually don't want to delete (like when you change folders,
> IIRC) just hit return twice (once when you find and choose your mailbox
> and once to take the default "no").
That seems like such a hack, and is awfully inconvenient. How do most
other mutt users handle deleting and purging messages? Does anyone else
actually use delete=no?
Paul
