winans@sirius.mcs.anl.gov (John Winans) (04/13/91)
In article <1991Apr13.100211.5937@pegasus.com> tony@pegasus.com (Anthony Hoelscher) writes: >I have a potentially dumb question. > >I've already seen a discussion here about implenting TCP/IP on Coherent and >noticed the comment that for now it's a dream due to Coherent's 64K. Remember that Coherent allows 64K code & 64K Data... The SMALL memory model... not the CP/M memory model ;-) >So does the inability to get TCP/IP working on Coherent also rule out the >possibility of running a SLIP type connection? I ask this since I don't >have a full understanding of SLIP -- does SLIP work independently of >TCP/IP or does it work on top of TCP/IP? > >thanks and aloha Aloha?? Very punny! SLIP sitts under TCP/IP so you need TCP/IP as well. If all you wanna do is support 1 (and only 1) socket/session (like to support a telnet or FTP link) You should be able to do that in Coherent pretty easily. In fact, since the Coherent Driver Development Kit comes with the source for an ARCNET driver, some of the grunge work is already done... if you can deal with ARCNET. I am probably going to join the group of people working on TCP/IP for Coherent. My first view on the project is positive because there are enough minor device numbers/device to write a multiplexed stream driver for something like an ethernet card. There is support for allocating segments of memory for use by the driver (don't know how many yet... probably just 1 64K chunk.) And I have seen UDP-level TCP/IP services supplied in less than 64K of code and 64K of data (see PCBRIDGE... an ethernet gateway package that runs under DOS) so... I think a streams version of TCP/IP can be done. Has anyone out there done messing around with the Driver Development Kit stuff yet? So far, I have only rummaged around thru the sources. --John -- ! John Winans Advanced Computing Research Facility ! ! winans@mcs.anl.gov Argonne National Laboratory, Illinois ! ! ! !"The large print giveth, and the small print taketh away"-- Tom Waits !