|
|
| GeoCommunity Mailing List |
| |
| Mailing List Archives |
| Subject: | Re: GISList: Re: Effective Standards |
| Date: |
04/24/2003 06:05:01 PM |
| From: |
Peter Baumann |
|
|
(I reply to this because it has a more suitable subject, discussion evolving
this way)
> The reason OGC standards are going nowhere is that most GIS users want fast,
> modern, effective GIS at very low cost (GIS users understandably want to be
> a part of the wonderful price/performance revolution that has swept almost
> all other areas of computer hardware and software)
From this argument and another one that came up (in essence, "give me freeware
implementations") I hear the desire for high quality at low/no price. How should
that work, aka economic perpetuum mobile? You'd argue that WWW came into the
Internet for free, but how much research was invested into getting it to this
crisp point? (Not just CERN, but also other research done prior & around). Has
been paid by the tax payer = us, and more than $1M+. Although I do believe in
open source software as a serious alternative to the classical vendor models I
don't see it as the Holy Grail per se.
> and using OGC is the
> opposite of that - it's a formula for slow, overpriced, inefficient,
> bureaucratic products that don't make sense for most users.
Maybe I miss something in my technical understanding, but up to now I couldn't
find that. Maybe we're just getting heated.
> Setting aside those examples not relevant to how high tech standards emerge
> (screws), let's not re-write history: for the most part those standards that
> populate our tech world were rapidly adapted, at times quite literally
> overnight, because they made sense and were very convenient at the time for
> achieving maximum price/performance. Centronics parallel ports, for
> example, were universally adopted by the printer industry within a few
> months, as were the various standards surrounding the Wintel clone
> architecture.
As for Centronics: do you speak of the time from publication to adoption, or
from setting up specs, doing prototype implementations, boiling down results
into generally available literature, and then to adoption? There has been a lot
of hidden effort until the standards proposals emerged.
As for Wintel: Nice example - there was a host of small computers before, from
the VAX PDP-11 to the microprocessors, running CP/M, DR-DOS, PC-DOS, not to
speak of Apple MacIntosh. Only in the very end, when market success was mighty
clear, IBM threw "The PC" into the market, BTW setting a de facto standard that
was certainly not state of the art (and actually not meant as that by the guys
that originally built the IBM PC). So we find (i) long development with a lot of
"waste" (cynical for companies & technology being kicked out) and (ii) a "de
facto standard" set by shere market power of the (then) #1. Shall we declare
ArcGis as the standard? (Flames on all but this please - I don't want to imply
anything on ESRI products with this statement, it's just an arbitrary example.)
> If OpenGIS wrote a binary client/server specification that included a
> BSD-licensed reference implementation library it would be immediately
> (a) useful and (b) used.
Why then isn't the U of Minnesota server the gold standard, used by all?
(Again, just an example...)
I can't quite follow what a "binary spec" is, so maybe I'm way off topic with
what comes now: Generally speaking, I trust in state of the art software
engineering where we have abstract functionality specs + interfaces using
model-based specification techniques (hopefully with some automatic consistency
checks) that become manifest in this or that concrete implementation. For
example, specify in UML and then break it down to XML.
Which brings me back to an interesting question, seriously: what concrete
alternative do you see to OGC and ISO? ("open source" is a task, not a
solution... :-)
Regards,
Peter
PS:
> [I apologize for piling on against OGC, but this lame excuse of "standards
> are so hard to establish" in lieu of touching base with the reality of OGC's
> leaden unpopularity irked me into penning this missive...]
Oops, (i) this was not meant as an excuse, and (ii) at least here in Europe I
realise a steadily growing popularity of OGC.
PS/2:
> All this hooey about XML/GML being "human
> readable" is irrelevant in the presence of a useable (and implemented)
> API. If you leave the implentation of things solely to vendors they will
> botch it up (accidentally or on purpose), and defeat your aims.
About XML I agree completely. A similar superstition was laid aside decades ago
by the database guys who had thought they had a "structured ENGLISH query
language". And SQL is much more readable than XML. IMO XML is the assembler that
does its job ni
|
|
Sponsored by:
For information
regarding
advertising rates
Click Here!
|
