RE: [cgmopen-members] Fwd: Re: CGM Interested parties

[martin.jackson@notes.canadair.ca]

Lofton,

You are right it should not be neccesary to implement style guides to
enable cgm interchange - but realistically I think it is going to remain
neccesary. All we can hope for is to limit the need.

In the past few years we have insisted that our suppliers only use certain
approved graphic software. This has been a key strategy in reducing
transfer errors. Even then some vendors simply refuse to purchase the
required software. We then rework the data to standard - we have no choice
- without reliable master data we are sunk. The only alternative would be
to rastorise.  This does not make sense particularly since one of the
greatest values of CGM to us is the ability to reuse graphic components
over and over again - time savings are huge. Allthough ATA and NIST
certification are contractual requirements it is often impossible to get
some companies to spend a relativley minor amount of money on the neccesary
software. As a corporation we cannot do more.

Despite this 90% of our business partners have moved over to use our list
of approved software. Things are getting much better!

Even though we now have most of our suppliers providing conforming CGMs it
is still essential to control the data creation via a style guide. For
example a dashed line is perfectly legal in several separate lines - but
useless for reusability. If a BSpline is converted to a polyline its
reusability is lost. It is a fact that many suppliers have tried to work
around the CGM conformance requirement by purchasing one licence then batch
converting graphics from other sources such as Illustrator or CAD
applications. What do you get - useless, often huge,  syntactically correct
graphics. In one case a supplier used a rastor to vector convertor and then
ran it through an approved illustration tool. The result; a technically
correct, 3Meg, unreadable, unusable graphic. This supplier had a one man
publications group comprised of an engineer wondering what he was doing
this stuff for!

I believe what is needed is ATA exchange rules that refine and narrow down
Generator and Interpretor rules to address the basic goodness issue of
preserving geometric intelligence. In some cases this may require not
supporting some existing attributes. Delivery is less of an issue so long
as the source CGM is consistent and reliable.