16 May
2025
16 May
'25
6:22 p.m.
On Friday, May 16th, 2025 at 10:58 AM, Clem Cole <clemc(a)ccc.com> wrote:
below.
On Fri, May 16, 2025 at 12:01 PM segaloco via TUHS <tuhs(a)tuhs.org> wrote:
Insightful as always Clem, thanks for the background. To me that sounds more
like AT&T seeing the horse escaping the barn and trying to ensure they're a
relevant part of the conversation, more for the sake of marketing their system
and cornering things rather than a good faith attempt to improve and extend the
ideas being set forth by the /usr/group effort. Not to discount the good
intentions of plenty of folks involved, but that certainly sounds more like a
case of "shoot, we should be the arbiter of this information" for the sake of
their bottom line. My interpretation anyway.
Aside but I recently landed a copy of SVID Issue 1, previously I only had
Issue 3 (SVR4 era blue books, 5 volumes) and had seen auctions for Issue 2
that I didn't act on. Issue 1 is published in a nice hard cover, has a more
minimalist perspective pattern on the cover than Issue 2 and is only one
volume covering the C interface while mentioning future extension efforts
such as basic utilities and the like.
- Matt G.
P.S. While on the subject of SVID, just throwing out there that I have seen
auctions for a "Volume VI" in the Issue 3 era. No such volume is referenced
in any of the others and the one auction I jumped on never arrived. When
pressed on it the seller simply gave me a refund. This has all amounted to
doubt that such a volume actually exists, but I've seen at least one grainy
photo (because book sellers can't be bothered to show adequate pictures of
the books they're selling but expect $$$ anyway) that appeared to show a
cover saying "Volume VI" but in typical book seller fashion, it was far too
grainy and low resolution to make out any details. In any case, I share
this in the hopes someone might be familiar with what I'm talking about and
can confirm/deny the existence of a 6th volume in this set.
I'm curious if anyone has the scoop on this.
To my knowledge the 1984
/usr/group standard constitutes the earliest attempt at a vendor-neutral UNIX
standard. AT&T then comes along in 1985 with the first issue of the SVID, based
largely on SVR2 from what I know.
There was a huge marketing campaign, "System V. Consider it Standard." But the
>>users<<, particularly those weaned on BSD, said "hardly."
/usr/group was an attempt to deal with Ultrix, HP-UX, AIX, and, much less, Sys III/V.
SVID came later, and it was an attempt to force it down people's throats.
The AT&T folks were sometimes a tad nasty at the POSIX meeting and wanted IEEE to
"just use it," and we say, "no. It's incomplete and just plain wrong
is so many places." The whole tar/cpio stuff from /usr/group was a great example of
the start of it, but even things like trying to define a directory entry was strained.
SVID did not have the new UCB directory system calls. For example, we all were certain
that if we ever had a different FS, we needed to remove physical formats from the
specification. There were no sockets, and yet nearly 100% of the working networking code
in the wild, including on MS-DOS, was using sockets.
The problem was that several people who came to the POSIX meetings post-SVID from
AT&T were from marketing and sales. At the same time, the core of the original
/usr/group and later POSIX teams were mostly engineering types. The sales/mktg folks were
trying to establish a brand, the engineers were trying to solve an issue were we had code
that did not work between our different systems.
ᐧ