[comp.emacs] Problems with large RMAIL files?

walters@chance.uucp (Chris Walters) (08/24/89)

We are experiencing strange behavior with RMAIL.  Messages in
large files (folders) appear to be truncated, and new messages
are inserted in strange places on occasion.  Once, a large
number of NULLS were inserted into the folder, where the rest of
the message should have been. Talk about junk mail!  Usually, I
bring in new messages into a new RMAIL buffer, and then ship
them out to other folders using the 'o' command.

I'm not a lisp hacker, so any insight would be appreciated.


 -- Chris Walters
    MITRE McLean Software Engineering Laboratory
 	{uunet,...}!gateway!community-chest!walters
	walters@community-chest.mitre.org

montnaro@sprite.crd.ge.com (Skip Montanaro) (08/28/89)

In article <65300@linus.UUCP> walters@chance.uucp (Chris Walters) writes:

   We are experiencing strange behavior with RMAIL.  Messages in large files
   (folders) appear to be truncated, and new messages are inserted in
   strange places on occasion.  Once, a large number of NULLS were inserted
   into the folder, where the rest of the message should have been.

My guess would be that you are using NFS, and mounting your read-write file
systems using the "soft" attribute. Mount them "hard" instead. 'Tis not
Emacs's problem, unless there is an error being returned on the write(2)
that it's not paying attention to.

--
Skip Montanaro (montanaro@sprite.crd.ge.com)

jthomas@nmsu.EDU (08/28/89)

chris> We are experiencing strange behavior with RMAIL.  Messages in
chris> large files (folders) appear to be truncated, and new messages
chris> are inserted in strange places on occasion.  Once, a large
chris> number of NULLS were inserted into the folder, where the rest of
chris> the message should have been. Talk about junk mail!  Usually, I
chris> bring in new messages into a new RMAIL buffer, and then ship
chris> them out to other folders using the 'o' command.

chris> I'm not a lisp hacker, so any insight would be appreciated.

You don't say what kind of machine/OS/file system you are using.  Please
do.

I've seen that effect under SunOS 3.2 and 3.4 under NFS with gnu emacs
18.51 .  I haven't had any large files recently under 3.5 / haven't had the
problem any more (your guess is as good as mine ::-{(} ).

What size is "large"?  My problems seemed to start at ~600K .  Are the
blocks of nulls in multiples of disk blocks ( e.g., 512 bytes)?  Do they
start at disk block boundaries?

Jim Thomas        jthomas@nmsu.edu