www.delorie.com/archives/browse.cgi   search  
Mail Archives: djgpp/2000/06/02/19:00:19.1

From: Richard Dawe <richdawe AT bigfoot DOT com>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: netbd.h, sockets.h, in.h and types.h don't work
Date: Fri, 02 Jun 2000 19:43:20 +0100
Organization: Customer of Planet Online
Lines: 83
Message-ID: <39380048.CD366DDF@bigfoot.com>
References: <Pine DOT SUN DOT 3 DOT 91 DOT 1000601085256 DOT 17554C AT is> <3936DA62 DOT 9581F9CE AT bigfoot DOT com> <3936feb4 DOT 2522956 AT news3 DOT banet DOT net>
NNTP-Posting-Host: modem-54.potassium.dialup.pol.co.uk
Mime-Version: 1.0
X-Trace: newsg4.svr.pol.co.uk 959986654 8158 62.136.18.54 (2 Jun 2000 22:57:34 GMT)
NNTP-Posting-Date: 2 Jun 2000 22:57:34 GMT
X-Complaints-To: abuse AT theplanet DOT net
X-Mailer: Mozilla 4.51 [en] (X11; I; Linux 2.2.14 i586)
X-Accept-Language: de,fr
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Hello.

"Peter J. Farley III" wrote:
> PMFJI here fellas

Whoa, what does "PMFJI" mean? Peter M. Farley Jumps In? ;)

> I ran into this myself recently with an open-source FSM (finite state
> machine) application generator called libero <http://www.imatix.com>. 
> I, too, got the "ntohl redefined" error (among others), and thus
> discovered this problem myself.

I forget - why does this error appear? I looked back at the reason this
thread started and I don't see why this happened. It looked to me like the
originator was using stale headers from an old version of libsocket, but
that wasn't confirmed. (I just read section 8.2 of the FAQ too, BTW).

BTW Eli, you refer to libsocket as "libsock" in a couple of places in the
FAQ. Also there seem to be some links to my old site:

http://www.geocities.com/SiliconValley/Lab/3216/

libmslot is one example. You can translate these URLs to the new location
by changing the above prefix to:

http://www.phekda.freeserve.co.uk/richdawe/

> But it really would be finer if the zip-picker asked if you wanted DOS
> networking capabilities,

Yes, there is agreement that the zip picker needs a couple of items added
to it.

> For those new to the networking libraries/headers, it would be nice to
> see a short FAQ somewhere on what libraries are needed for what
> networking application areas (i.e., do I need libnet AND WAT-TCP AND
> libsocket, or just two of them, or just one?  How do I choose?  Are
> there incompatibilities?  In what order should they be installed?
> Etc.)

OK, here's a short FAQ, off the top of my head:

. Watt-32 is for networking under pure DOS or under Windows with an NDIS
packet driver installed (ndis3pkt). It does BSD sockets, plus the WATTCP
API (from which it's derived).

. libsocket is for networking under Windows 3.x (untested for a long
time), Windows '95, possibly '98 (untested by me, reported to work/not
work), Windows NT (Unix domain sockets only, no TCP/IP, untested).
libsocket is a BSD sockets-only library.

. libnet works on DOS, Windows '95 (Winsock 1.x only). It has its own
APIs, protocols, so definitely does not do BSD sockets (*).

(*) I apologise that my point of view is polarised to BSD sockets. It's
what I'm used to. ;)

I think Watt-32 and libsocket will clash, since they both include the same
BSD socket networking headers (well, actually Watt-32 includes more). I
don't know enough libnet to commet here (perhaps Damien Yerrick can help
here? ;) ). I imagine it will not clash - since it has its own APIs, it
would seem sensible not to make it clash with BSD sockets.

So, you only need one. Now to the crunch: which one you choose? If you
want BSD sockets under pure DOS, then there is no choice: Watt-32. If your
programs are going to run under Windows '95, '98, then you could also
consider libsocket. I'm trying to think of an advantage here over Watt-32,
but I can't - hmmm, maybe: possibly smaller executables, since it's a thin
layer over Windows VxDs; copious amounts of documentation;
auto-configuration on Ethernets, means no fiddling with config files
(although you may have to do this).

This is my honest opinion, although it's pretty damning for libsocket. :(
Watt-32 is probably more reliable than libsocket by a significant margin.
libsocket development has been stymied by several bugs in the VxD from the
Coda project and a lack of time/enthusiasm on my part (sorry about the
enthusiasm bit - I think it's back now).

Bye,

-- 
Richard Dawe
richdawe AT bigfoot DOT com ICQ 47595498 http://www.bigfoot.com/~richdawe/

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019