Newsgroups: comp.unix.programmer
Date: 27 Jul 1995 18:34:26 GMT
Organization: U S WEST Advanced Technologies
> socket in the same process? In other words, can system
> call such as select, read and write work with both
> types of protocol? Does Solaris 2.4 support such
> scenario?
>
> I run a process that has X.25 and TCP/IP sockets in it.
> The socket semantics are extremely similar, but
> seperate drivers operate on seperate address 'domains' in the
> kernel, so there are no first-order problems. This is
> in SunOs 4.1.3 (using SunNet X25 7.0.1), so it should work
> even better under Solaris 2.4 (although, you need the
> SunNet X25 8.x product to build the sockets into your kernel).
> It seems that TCP/IP comes free (INET domain), but X25 is extra.
> There may be some confusion when it comes to "out-of-band"
> signals and interrupts (who answers the call?), but only
> if things get too feature or bug laden.
If someone has gotten X.25 sockets to work in Solaris 2.4, I'd
sure like to talk to them. After beating my head into the sparse
Sockets appendix in the Sunlink X.25 8.0.1 Programmers reference,
only to receive intransigent 'Protocol not supported' errors,
I decided that maybe the "you must use /usr/ucb/cc" text in App. C meant
that gcc wouldn't work.
I eventually recoded my sockets-based X.25 library using System V
streams, which work OK (at least when transmitting data in
chunks smaller than a frame size) but which required a painful
learning curve when I was already deaing with enough Sun->Solaris
issues to fill my day with ample manual-flipping, hair-ripping
diversions.
---
Free Complaints!
#include <std_disclaim.h>
--------------------------------------------
I did get all of those problems with SunNetX.25 until I
found out where the list of Sun patches was from a local
sysadmin guy here. Some details:
-- It took 3 patches, with from 15 to 75
bugfixes in each, to get the SunNet going
-- None were mentioned or included in the
software delivered, even though they had
been around for quite a while
-- There was a K+R macro vs. ANSI macro issue
in the sys files with modern compilers
that mapped the ioctl() calls way off course,
and produced a bunch of
operation not supported.....
protocol not supported....
complaints,
so I had to use a -traditional flag with
the compiler
-- (other traumas I have forgotten)
--- the main thing was knowing someone who had
experience steering through the galaxy of
Sun patches. Sytem Adminstrators know these
things all too well.
--------hope this helps someone-------------