[comp.sources.bugs] Reporting bugs with mush..

dheller@cory.Berkeley.EDU (Dan Heller) (05/12/89)

A request to those posting "bugs" about mush.  Because Mush
runs under many different MTAs (sendmail, mmdf, delivermail, execmail,
smail, etc..), the folder format or protocol your site uses may be
vastly different from what I use.  If you are having a problem and
suspect it to be MTA specific, please be sure to mention which MTA you
use.  Also whether you use DOT_LOCK, MSG_SEPARATOR, would help.  It could
change my perspective on the problem and come to a quicker answer.

If you are having user interface problems, specify which user interface
you're talking about.  If you're using curses mode, then the problem isn't
likely to exist in the suntools mode.  And no matter which mode you use,
if you can test if the problem exists in the tty mode, that would help
as well.

Always try to mention what system you use -- system V, BSD, 4.3 or 4.2,
sun, risc machine, etc...

So far, few bugs are actually being reported -- there are more inquisitive
comments than anything else at this point.

BTW, those using NFS -- you should be sure that filesystems mounted
read/write should always be mounted "hard" -- if you mount an NFS filesystem
for read/write as "soft" and try to open a file for writing (e.g. an editor
or updating a folder), if the RPC times out and NFS returns a time-out error,
the file is -lost- and there is no way to recover it.  this is documented
in the NFS notes in some obscure sun manual somewhere.  I urge you to read
it or query a guru (please, not me).  If you want to allow a more legitimate
return from a timeout, then you can set the timeout value for a "reasonable"
packet timeout (limiting packet size helps as well).

Also note that a filesystem mounted for read only is best mounted as "soft".

If anyone questions this or refutes it, please don't discuss it -here-!
Mail me personally (but only to flame or argue -- no questions :-)
Dan Heller	<island!argv@sun.com>