stanonik@nprdc.arpa (Ron Stanonik) (08/23/85)
The 4.2bsd tftp implementation apparently doesn't properly handle "new lines" in netascii mode; ie, it just moves blocks of uninterpreted data, no different than octet mode. That's fine between similar systems, but failed between a pc and our vax. The tftp rfc refers to the ascii modifications in the telnet rfc; ie, "new lines" should be converted to <cr><lf> when sent and <cr><lf> should be converted to "new lines" when received. <cr> should be converted to <cr><nul> when when sent and <cr><nul> should be converted to <cr> when received. So, has anyone fixed(?) the 4.2bsd tftp? Thanks, Ron Stanonik stanonik@nprdc
guyton%condor@rand-unix.ARPA (08/23/85)
I've added netascii (and many other bug fixes) to the 4.2bsd
tftp user and server code. Since the conversion to/from
netascii slows the transfer down somewhat, I'm also adding in
a little read-ahead, write-behind logic to try and speed it
back up.
I've not (yet) asked Berkeley if I can just post this code;
so in the meantime I'll send it to anyone that has a 4.2bsd
source license (and sends me a copy of it along with their
network address).
-- Jim Guyton
arpa: guyton@rand-unix.arpa
uucp: {decvax, sdcrdcf, vortex}!randvax!guyton
post: Rand Corporation, 1700 Main Street, Santa Monica, CA 90406