[comp.mail.mh] Help! slocal doesn't work on Sun

bd@hpsemc.UUCP (bob desinger) (11/24/87)

> We are trying to get MH 6.5 working on a Sun 2/170 ...
> Here is the problem (memory fault on slocal). What is .prog
> doing anyway? 
>
> /usr/sharat/.forward: line 1: sharat... forwarding to "| /usr/lib/mh/slocal -
verbose  -user sharat"
> /usr/sharat/.forward: line 2: sharat... forwarding to \sharat
> "| /usr/lib/mh/slocal -verbose  -user sharat"... Connecting to .prog...
> sh: 1792 Memory fault
> "| /usr/lib/mh/slocal -verbose  -user sharat"... unknown mailer error 139
> \sharat... Connecting to .local...
> \sharat... Sent

Sendmail prints ".prog" to tell you that it knows you're mailing to a
program (from the leading `|' in the alias).  I'm not sure if that's what
you're asking, though.

If I was debugging it, I'd check the arguments to slocal for
correctness, and check the permissions on directories from / down to
slocal (to ensure that Sendmail can get to them).  Sendmail is quite
picky about file permissions, too---for example, mailing to a file
requires that the file be mode 0666 [or at least that restriction used
to be important]---so perhaps Sendmail doesn't like the permissions.
On the other hand, I wouldn't expect a memory fault to arise out of
mode problems.  There should be a "Sendmail Installation and
Configuration" document somewhere among the system admin papers that
came with your system; check that out.

Hope this gets you started,
bob desinger

ted@BRAGGVAX.ARPA (11/25/87)

I've never played with slocal, but it's been my experience that whenever
some part of MH doesn't work on a Sun, there is a pointer problem of some
kind, usually dereferencing null pointers.

				Ted Nolan
				ted@braggvax.arpa

sow@cad.luth.se (Sven-Ove Westberg) (11/25/87)

> > We are trying to get MH 6.5 working on a Sun 2/170 ...
> > Here is the problem (memory fault on slocal). What is .prog
> > doing anyway? 
> >
> > /usr/sharat/.forward: line 1: sharat... forwarding to "| /usr/lib/mh/slocal -
> verbose  -user sharat"
> > /usr/sharat/.forward: line 2: sharat... forwarding to \sharat
> > "| /usr/lib/mh/slocal -verbose  -user sharat"... Connecting to .prog...
> > sh: 1792 Memory fault
> > "| /usr/lib/mh/slocal -verbose  -user sharat"... unknown mailer error 139
> > \sharat... Connecting to .local...
> > \sharat... Sent

> If I was debugging it, I'd check the arguments to slocal for
> correctness, and check the permissions on directories from / down to

No, the best way is to send a mail WITHOUT using slocal. 
Then copy your mailbox to foo and debug slocal separated from sendmail. 

/usr/lib/mh/slocal -verbose -user sharat < foo

This is also a good way to test a new maildelivery file.
If this works try to use it with sendmail.

My sendmail dislikes ALL flags to slocal. I don't why and 
I'm running MH6.4 so I don't have any manual page to slocal
describing all the options.  If anyone have a manual page to
slocal I would be pleased if someone could send it to me.

About the bugs it was "some :-)" in the slocal, rcvstore and rcvtty.
But it was 1 year since I fixed them in 6.4 NOT 6.5. I didn't have
access to 6.5 sources. I asked for the diffs between the MH distributed
with bsd4.3 and MH6.5, but all I got was nilpointers :-)

Sven-Ove Westberg, CAD, University of Lulea, S-951 87 Lulea, Sweden.
UUCP:    {uunet,mcvax}!enea!cad.luth.se!sow
Internet: sow@cad.luth.se

goldberg@russell.STANFORD.EDU (Jeffrey Goldberg) (11/25/87)

In article <8711242334.aa05119@ICS.UCI.EDU> bd@hpsemc.UUCP (bob desinger) writes:

[someone pointing out that slocal dumps core on a Sun 2/170]

>If I was debugging it, I'd check the arguments to slocal for
>correctness, and check the permissions on directories from / down to
>slocal (to ensure that Sendmail can get to them).  Sendmail is quite
>picky about file permissions, too---for example, mailing to a file
>requires that the file be mode 0666 [or at least that restriction used
>to be important]---so perhaps Sendmail doesn't like the permissions.
>On the other hand, I wouldn't expect a memory fault to arise out of
>mode problems.

>Hope this gets you started,
>bob desinger

I wasn't the one who posted the original question but I have the
same problem.  The problem is not with the sendmail/slocal
interaction; so there is no need to go into thw bowels of sendmail.

I know this because I fed slocal a legal mail message by hand, and
it still bombed.

A few months ago I posted the same question, for a Sun 3/280S and
every Sun users who knew what slocal was for reported the same
problem with MH 6.4 and 6.5.

One person told be that they got it to work by recompiling all of
MH with the -g compiler flag.  I haven't done that yet, though I
did try it with the -go flag (I wanted to save on space).  It did
not work.  I will try recompiling with the -g flag anyday now.

-jeff goldberg


-- 
Jeff Goldberg         Internet: goldberg@russell.stanford.edu