: > : Since I've been on the Internet and finally figured out the newsgroup
: > : thing (about two week ago), I've been keeping an eye on this newsgroup
: > : and haven't seen anything come through.
: > :
: > I suspect that the lack of postings lately has something to do with
: > the holiday season.
: > : It might be a good idea if everyone that has this newsgroup in their
: > : list to make a posting so everyone else can get a idea of the audience.
Just reading is far too reasonable an idea to expect action. What is peculiar
about reading newsnet is that all readers have equal access as speakers.
: I suspect that the PICK database market is still quite diverse and this
: is reflected in the various discussions that take place.
: I only really partake in specific discussions of Prime INFORMATION,
: uniVerse, PI/open and UniData (and related products) myself, although
: I find the technical discussions on the PICK Operating System interesting
: at times. But I'm a little prejudiced and find it still a little
: backwards, also (strangely) quite forward in some areas.
If you mean that 'traditional Pick' is backwards as respects Prime, but
of course. As Rod Burns said, "We wanted to be able to do all the
things we couldn't at Microdata." If you mean that Pick and Prime, to
include the 'traditional' and 'modern' (? :-)) implements, as opposed
to the rest of the 'things' (OSs, DBs, 3GLs, 4GLs whatever) out there,
that's a more interesting question.
: It provides the goods. Always has...maybe now it showing it's age?
Well, it's problem has always been that it didn't _act_ its age.
'Showing its age' is a bit more problematic. One of it's 'charms'
is not of a particular phase of development of computers. This
appears to be why it seems to be 'one of those' for the database
flavor of the half decade -- for each half decade. If, on the other
hand, you mean that 'obvious' extentions do not seem to have been
made in a timely fashion -- you bet.
: I predict that it may wither if the various PICK database vendors stop
: kidding themselves they don't exist and set some real standards.
By which I take you to mean that the continued vitality is due to the
lack of centrally set standards. I agree. :-) I suspect however that
you actually meant that some real standards _ought_ to be set. At this
point, that would mean Pick, Vmark, Unidata, maybe Univision and Jbase.
Now, this sounds even worse that the SMA standards effort, which started
in September of '84. The _major_ breakthrough was the documentation
of the FS f-correlative wherein the inequalities are in the same
order of execution as subraction, all in the order of the Pick R77.
That took two years, and the offer of the SMA technical standards
organization to resign en masse. Smirk. Other big decisions were
the adoption of the Microdata Locate and the Schellenbach Execute.
But I remenisce. It's my actual experience that, if one vendor
does something useful, it gets adopted in due time by others. But
the real problem is that very subtle behaviours, things which the
original implementors didn't even realize, on occasion, need to
be have the same functionality. I note the case of the 'R00N'
mask in R83 as an input conversion, used in basic, didn't work
quite the same way at Microdata.... So, what is R00N? Doesn't
matter. Gotta work the same way. How's that? Not like this...
Standardization at this level does not happen without DARPA budgets.
Also are interesting is the problem of upward compatibility, wherein
upgrading from one release to the next on the same machine makes
things stop working (horrors.... couldn't happen, could it?), and
worse, the problem of bacwards comptiblity: You are not allowed
to add anything which didn't work in the past, lest applications
can't be run on previous machines. Blessedly, _those_ people are
off to a different part of the computer business.
: More importantly new development tools are needed now. And I'm not
: talking the likes of System Builder/ SB+ or TOADS or CuBIC, etc.,
: etc. Just more basic stuff...
What do you mean: More basic stuff is needed, or that more SB+ is just
basic stuff? Either hypothesis is interesting.
: Because I'm Prime INFORMATION (PI) bias, let us start from what it offers
: (what) both UniData and uniVerse copied and in some cases improved):
: - General (generic) Terminal Interface;
: - Command-line Editing (also from within BASIC INPUT) with programmable
: - General Calling Interface (GCI) Standard needs to be set. Calling
external 3GL languages;
: - Information Calling Interface (ICI) accessing database from an
external 3GL language or source;
: - Indexed Online Help system;
All above, yes, particularly online help. There's no reason why
it wasn't there all those years.
: - UPDATE.RECORD verb to update via Dictionary names;
By which you mean....? Update validation filter? This is part of a
discusson about, "What is the Pick database, really, and what should
it be used for, and expected to do?"
: - SHOW verb to provide pagination with LIST-like functionality;
What's a SHOW verb?
: - Phase-out Correlatives! These are just not necessary! Surely
there is somebody out there whom already has a translation
: - PROC must die! Get real! Translator anybody? Surely if they
can write an interperter they can write a translator??? :)
Really, now, why? Some of us are in the habit of generating correlatives.
And I find PQ proc as a jobstream language preferable to basic. I do not
think that it should be used as an application definition language, ala
PQN. But some disagree with me on that. Of course (most) of it can be
translated to basic. And all _sorts_ of things won't work any more. :-)
Interestingly, it's quite difficult to build an emulator for the full
correlative capability. This comes to light, since F-correlatives
could not be run from basic until quite late. Therefore, everyone
attempted to write translators and interpretors. They didn't tend
to work. And _then_ you get to the subtle differences between
R77 and 3.2... Which are still subtle.
: I realise that many of the advanced PICK O/S incorporate many of
: these features, but they of course are all propertiary (just like
: Prime INFORMATION's).
: There are no standards - they tried and failed before (?). Time for
: another go?
Well, they worked just fine, as far as they went. There was not a
motiviation to go far with the standards. Each of the vendors though
that they could get more of an advantage over the other with its own
development, rather than by expanding the market by making the machines
'similar'. Just like everone else. What you won't find, I think,
is any analogous case of a large piece of software, with this many
vendors for this long a time anywhere else in the computer business.
I would also argue, that for all its lack of standards, the degree
of compatibility exceeds that of contempraneous Unix machines. I
also agree that something like a POSIX compatibility level would be
Thanks for your comment. What _I_ want is an order of 1 access file
for my PC, so that I can implement English (with labeled storage and
looping and brancing in Fcorrelatives, so that I can build a Turing
machine) and VisualData/Basic++... We could call WeWe all the way home.
And then I want to bind OLE2 objects into Fcorrelatives.... all with
the intent of implementing a throwaway code machine....