7 Sep
2018
7 Sep
'18
12:22 a.m.
On Wed, Sep 5, 2018 at 10:12 PM Ed Carp <erc(a)pobox.com> wrote:
I've read that Taylor UUCP fixed a bunch of
issues with HDB UUCP -
have you considered using it?
Hmmm.. the only thing it really fixed was the licensing. HBD was part of
distributed in the 'toolkit' and as part of Svs V. I think it was part of
PWB 3.0, as we shipped it with RTU; but we have have gotten it via the
toolkit license [I'm too lazy to look at Warren's System files to check].
HDB was the huge fix of the original version that went 'wide' with V7 and
BSD. The primary changes were in house directories and the queues were
handled, which had huge performance impact; particularly for large sites
with an heavy UUCPnet load (a.k.a. the 'Usenet'). The other things that
was in HDB was some collected alternate protocols besides Greg Chesson's
original 'g' protocol; although those had all been distributed on the
Usenet as net.noise, so all it really did is become a packaging thing.
There were a number of UUCP clones written in those days, and PC-uucp was
popular for CP/M and later DOS systems that wanted to join the Usenet [Rick
Adams, I think had a packaged version of some of them for his non-UNIX
customers IIRC).
As to why Dave Taylor decided to write another one in thoses, you'ld have
to ask him; but in the end it replaced the V7 one in BSD. But a lot of
effort was made to make sure Taylor UUCP was more than a functional
work-alike. It had admin like HBD and use the same queues etc. What I
do not remember, Mary Ann might, is if Berkeley had been running HDB from
the Toolkit on ucbvax internally, but could not have easily redistribute it
( seem to remember there were). They funny part is that most Academics
had a toolkit license by then becsuse they wanted HDB and ksh as a minimum,
so most were running HDB; but the rules in the toolkit for academics were
different than the original base license (I've forgotten them -- I had left
academia so it did not effect me). But it was around this time the Keith
was start to try to remove code that was knows to have AT&T copyright
issues, when their was an alternative implementation that had BSD style
licenses (which Taylor UUCP, as did Clark's troff as I recall).
BTW Grant, Linux picked up Taylor UUCP after BSD (it was not in the
original distros and I suspect would have crashed the kernel in those days
if the site was as all loaded).
ᐧ