Re: flexfax: sendpage+hfaxd, SNPP DATA broken? (4.0pl1 Sol2.5binary)

["Thomas A. Szybist" <szybist@boxhill.com>]

In message <346722A6.32F389DC@memphis.edu>, Mike Frohme writes:
> Matthias.Apitz@SOFTCON.de wrote:
> > Mike Frohme wrote:
> >     Matthias.Apitz@SOFTCON.de wrote:
> >     > Mike Frohme wrote:
> >     >
> >     >     Hello,
> >     >
> >     >         Just picked up HylaFax 4.0pl1 (Solaris 2.5, binary dist
> >     >         from SGI) and have installed.  Fax features work great,
> >     >         but I'm having minor problems w/Alpha paging:
> >     >
> >     >         sendpage works great when you specify the message in the
> >     >         command line.  Specifying the message via STDIN fails - the
> >     >         read never completes.
> >     >
> >     >         It appears that the MESS command is AOK, but the DATA command
> >     >         hangs in this distribution.
> >     >
> >     >     telnet localhost 444
> >     >     Trying 127.0.0.1...
> >     >     Connected to localhost.
> >     >     Escape character is '^]'.
> >     >     220 palantir.memphis.edu SNPP server (HylaFAX (tm) Version 4.0pl1)
> >     >     ready.
> >     >     login XX
> >     >     230 User XX logged in.
> >     >     page frohme
> >     >     250 Pager ID accepted; provider: 9,XXXXXXX pin: XXXXXXX jobid: 35.
> >     >     DATA
> >     >     354 Begin message input; end with <CRLF>'.'<CRLF>.
> >     >     Test message
> >     >     Line #2
> >     > [perhaps here was a line containing only a dot "." and the MTA
> >     >  cuts your mail here because the dot terminates SMTP also]
> >     >
> >         Yes, I manually entered a '.', but it doesn't work.  It *is*
> >         reading data on the port, because the server issues a
> >         550 Error, message too long; max 128 characters,  if I go
> >         over the limit.
> > 
> > Oh. This is another situation. The sendpage(1) failed
> > because the max length of the message is reached (and the DATA
> > command is aborted and hfaxd(1M) isn't waiting for a "." anymore).
> > Just say "quit" and hfaxd(1M) will terminate this session.
> > Does it also ignores the "<CRLF>.<CRLF>" if you don't reach the limit?
> > 
>         Yes, that's the point.  It *never* (as far as I can tell) sees
>         the "<CRLF>.<CRLF>" in DATA mode.
> 
>         I added the comments about "550 error" only to validate that the 
>         server is actually reading data from the channel.  Its just not
>         seeing the end-of-input.
> 
> -- 
> Mike Frohme                                     frohme@memphis.edu
> Information Systems - The University of Memphis       901/678-8820
> -- 
> Mike Frohme                                     frohme@memphis.edu
> Information Systems - The University of Memphis       901/678-8820

Hi,

I did manage to get a Solaris 2.5 machine that even had 
the binary distribution on it.  I was able to reproduce
the problem on that machine.  Truss on hfaxd showed a
read attempt after the ".".  Hmmm...  Then it justed polled.
Clearly it was missing the ".".  

I replaced hfaxd with what I compiled for Solaris 2.6 and 
that worked fine. The system call sequence looked pretty
much the same for both binaries.

All I can think of is that I compiled the 2.5 binaries with
gcc 2.7.2 and the 2.6 binary with 2.7.2.3.  I have heard
of some optimizer problems with early versions of 2.7.2.

Again, you can contact me if you would like to try this binary.

Tom Szybist
szybist@boxhill.com