by Peter Nurkse, Sun Microsystems

brought to your desktop by PTC/USER and PTC and Sun Microsystems.

This is the seventh year of this email newsletter from the PTC/USER International User Conferences, this time from Orlando, Florida. Same place as the first newsletter, in 1997, and also the 2000 newsletter.

As in previous years, views and opinions are just of the writer, and not of any organization or company. The purpose of the newsletter is still just to let people who weren't able to attend the conference participate in some of the daily activities as they happen. It can't replace coming to the conference and experiencing it yourself.

Contents

- PTC direction

- Product Development System

- Core Modeling

- Three Ages of the Pro/E GUI

- PTC direction

PTC seems to be embarking on a big fundamental change. Customers are usually concerned when a key supplier, like PTC, changes course. But we should all know that in these days and times change is inevitable.

The big change isn't a product, or a new product. PTC has done that before, several times even. PTC began with one product, Pro/E, and then developed separate modules, and then different products. And then with the acquisition of CV in 1997 PTC acquired Windchill and the basis for a whole new family of data management products.

So, products, products, products, that could be the PTC story to date. But now the company is planning to develop and deliver systems. Our familiar product names remain, of course, Pro/E and Intralink and the rest, but they will be developed and tested and delivered together as a system. And you can still buy and use any individual product or version you want, but PTC's focus has changed, from products to systems.

That is probably a more fundamental change than just adding products or mixing products, which is what PTC has largely done to date. So this could be the biggest change in the company's history, a change that will work itself out over years, this is just the beginning.

With any big plan like this, it's probably important to recognize the limitations of the plan. Design and data management will probably always be fundamentally different disciplines, appealing even to different people, different personalities. Design thrives on variation and spontaneity, while data management needs to lock things down.

Within PTC itself customers still see that difference. Ask the same question from someone from the Pro/E side of PTC and from someone from the Windchill side of PTC, and you can get two very different answers. And that may continue, because design people and data management people see things differently, even looking at the same thing.

The big plan, the new direction, to make and deliver systems, should over time solve any number of problems getting PTC products to work, and to work together. And it's a major competitive advantage for PTC, because nobody else seems to have such a comprehensive solution without using different packages which were never even designed to work together.

But design and data management are different enough that they will probably always remain creative opposites, inspiring and driving the system development. One system that combines both design and data mgmt. will have to recognize and respect the differences between them, to be effective. It'll be interesting times for PTC, balancing design and data mgmt. within a single system, and not imposing a data mgmt. solution on design or a design solution on data management.

- Product Development System

In the published program this talk by Jim Heppelman, PTC Chief Product Officer, was originally titled "Five Reasons NOT to Adopt Pro/E Wildfire!". Could be the title was just chosen to ensure a packed room, because Jim immediately abandoned it, saying that the subject has already been "beaten to death" on the PTC/USER email.

Dick Harrison couldn't make it to the meeting, feeling a bit ill at the airport in Boston before getting on the plane this morning, so Jim delivered that presentation in the afternoon too. That was more on the overall business, revenue and profits, but parts of it are included here.

Probably some people were disappointed not to delve into Wildfire details. But Jim did have bigger fish to fry, namely, Product Development Systems. And PTC is pretty confident about Wildfire overall direction (see the next story, on Core Modeling).

PTC representatives are often quite candid about quality. And Jim said that he thought Wildfire had the best quality of any Pro/E release in the last 5 years. Which would be about when PTC got Windchill, and Dick Harrison has told us before at these conferences how Windchill did distract PTC for some couple of years from Pro/E and Pro/E quality.

Jim did also mention, what probably most people know, that Wildfire did have more user involvement before release than any other version of Pro/E. However we can see that did set expectations, people did look to see their contributions included in the final product. Perhaps some way is needed so advance user testers can see what happened to their comments on the product. If things turned out differently that what you wanted, at least you'd probably like to know why.

Jim said that "PTC used to focus 100% on engineering", which was probably true, it was an engineering company. But engineers themselves know that a company that's 100% focused on engineering isn't a good company to do business with. You want more skills, marketing to help interpret requirements, services for support, even a helpful sales rep. So PTC has pulled back from that close focus on software engineering within PTC to get more involved in more areas.

Taking that kind of overall view to PTC and competing products, PTC found that product databases today are composed of tools and databases and "weak connections" between them. How things work together becomes the key question, not this or that feature here or there.

There's the system business that PTC is aiming for, "how things work together". There PTC see themselves different from a traditional Systems Integrator, like IBM or EDS. PTC would be a System Provider, delivering a complete system, but not building every system out of bits and pieces, like a traditional Systems Integrator. That seems a new phrase, System Provider, but that's good for PTC, it emphasizes PTC's strengths, having a relatively simple set of products to cover design and data mgmt. Alternatives from other companies require 2 or 3 or 6 or 7 different packages to get coverage in just one area (7 packages for EDS data mgmt.).

In case you haven't already noticed, Jim said that up to now PTC products have been developed in parallel, by entirely independent product teams, each team concerned to get the very best results for their products, and not necessarily for the other products.

To develop systems which integrate different products better, PTC will now include a System Planning stage at the beginning of each product cycle, so that all the products get some coordinated overall direction and input. And also a System Testing stage at the end of the cycle, where the products are tested together. That's new, up to now PTC product testing has been limited to checking the interfaces with other products, for each product, and has not included running everything together.

Already we can probably look forward a bit, and expect that PTC won't keep the product teams so separate, but may mix and match them, so then we might have a data mgmt. guy and a CAD guy working together on some common feature. Many combinations possible in the future, lots of potential.

For testing, PTC is going to develop and use a set of Certified Applied Practices, to keep the testing fixed on real world needs. Examples are Bottom-up Design, Top-down Design, Search and Re-use, Release to Mfg. Each a scenario of steps and stages involving different tools, both design and
data mgmt.

So what do these Product Development Systems look like, and when are they available? Here's a schedule, but Jim emphasized it all depends on John Vreeland, who will be in charge of the testing. If John says it isn't ready, it doesn't ship on the scheduled date. There's an interview with John in the Fall 2002 issue of Profiles magazine, available on-line at http://www.profilesmagazine.com/p21/interview.html. Here's the initial schedule:

PDS 1.0 6/2003 Pro/E 2001 2003010, Intralink 3.2 PDMlink 6.2.6 DSU4

PDS 2.0 7/2003 (Pro/E same) Intralink 3.3 PDMLink 6.2.6 DSU5

PDS 3.0 8/2003 Wildfire (Intralink same) (PDMLink same)

Nothing major so far, in terms of different products. But these first 3 system releases are priming the pump, getting ready for more. And the major point really is the mention of a system being delivered: that is major, up to now PTC has only delivered products.

You can see in these first 3 system releases a general PDS rule, and probably a good rule with any large system: change only one major component at a time, only one. That's a very different approach from the traditional product approach, which emphasizes releasing each product as
often as possible.

The next 3 releases are:

PDS 4.0 11/2003 (Wildfire same) (Intralink same) Windchill 7.0

PDS 5.0 4/2004 Pro/E next release (Intralink same) (Windchill same)

PDS 6.0 11/2004 (Pro/E same) (Intralink same) Windchill next release

So one big advantage here for customers is that you shouldn't have to face new versions of everything descending down on you all at the same time, just because every product team wanted to get out a new version at the same time. Now releases become system releases, and each product takes its turn, they wait for each other.

Jim emphasized products are still available a la carte. But he said you'd be best off to stay with the roadmap, or at least in sight of the roadmap. He gave due credit to us customers, saying that in the past a major difference for us was that we were doing all this system testing ourselves, running different products together. He also said in the afternoon that Intralink was going to be here "as far as I can see", and you can see that in the PDS major components: Intralink is a major component, equal with Pro/E and with Windchill in that if it changes that's a separate PDS release.

As a former Computervision employee himself, Jim may have been too modest to point out that PTC would not be able to deliver systems today if they had not bought up CV 6 years ago. Dick Harrison described at a past conference how PTC went about their due dilligence investigation of CV in Dec. 1977, and found "a present wrapped under the tree", namely, Windchill. Which CV was running as a skunk works in Minneapolis (hence the name), and which PTC didn't know about when they made the initial offer for CV.

Windchill became PTC's entrance into enterprise data management, and combined with CAD that is PTC's standing to be a system vendor now. Quite a Christmas present. PTC already releases CADDS5 in conjunction with Optegra, the CV data mgmt. product which goes back to 1983. Plus CV gave an example of a CAD company which became a data mgmt. company, and many CV people are within PTC today with that experience and background. Seems Computervision has made a bigger contribution to PTC than any other company, and that contribution continues as PTC develops the new role of a System Provider.

-- Core Modeling

Netesh Gohil covered this subject for PTC. On the subject of the topic of the day, Wildfire GUI, he was confident in the overall interface, although he said there was work to do too (see my essay on Ages of the Pro/E GUI following). He described the old menu manager approach as "hierarchical", and the new approach as not hierarchical. For example, with the old menu you had often to drill down a long way before you could start to do something. Now you can usually start right away (like, a hole), and see something, then modify it as you like. That's a key improvement to avoid huge dialog boxes filling up half the screen: you make choices as you need to, you don't have to face all the choices at once.

Netesh was positive all functionality is available, using the right mouse button and hot objects, or else collectors like Extrude Surfs and the dashboard. The goal for the dashboard is to show you the choices most people need for 80% of their work, and to give you the alternatives for the other 20%.

PTC found out that of spin/zoom/pan, spin was the most common function. So that became the middle mouse button by itself. Then zoom and pan are secondary. Netesh described the usual zoom in/out/in/out sequence. Which to any old Computervision user doesn't sound as efficient as the CV Drawing Window. That secondary smaller window let you zoom instantly throughout the visible model, in one area and then another, without ever doing a zoom out.

Netesh has a list of the 76 features in Pro/E that were developed over the years, with many different selection modes, surfaces and boundaries and so on. Now there's one selection method, and it doesn't depend on what mode you're in.

The dashboard can pause (becomes faint), which is where you can create construction geometry as needed.

Netesh thinks that making datums is largely a workaround for the pattern tool not handling rotations, that's why rotating and patterning is awkward. He wants to emphasize fixing the pattern tool problem, so that you just pick what you want to rotate, select an axis, and you're ready to go.

Insert menu always gives you a choice between object>action, or action>object. But when you edit an object, it's object>action only, because actions may depend on which object you select. For example, there are different copies for curves and for surfaces and for features (and the feature copy isn't in Wildfire yet, still the menu mgr.).

Patterns can be defined within any boundary you define on a plane, like perf hole patterns on a sheetmetal part. You can remove any individual hole instances in the pattern by digitizing the hole (it turns white), which is easier than editing holes in a table. The pattern can always be projected to a surface, but defining the boundary itself on a variable surface is a future.

View Manager handles simplified reps and explode states and display states and orientations in the one tool. And All States lets you create any combination of them. The goal here is "drawings as models", show any desired state in the model, without requiring a drawing. Model sections are due to be moved into the View Manager next.

Netesh gave an example of Wildfire efficiency in creating a feature: in the past, if you got an error, feature didn't intersect part, you'd have to cancel, then often create a surface (usually can always do that), then look for intersections, then create solid. With Wildfire, you can stop and check and convert between surfaces and solid, all within one feature command.

Rounds are just rounds: not simple or advanced, not a new set or a round set, but just rounds. You just start, and go to where you want (again avoiding a huge dialog box). There is a new round guaranteed to be machineable with a ball end cutter: before, Pro/E took some liberties using patches to define blends between rounds, which couldn't be machined with a ball end cutter.

Yellow is fixed now as the focus color, whatever you're looking at right now. Other entities visible will be shown fainter, always assuming a blue background.

About sketching first, before creating geometry: Netesh said that Pro/E requires everything (like a sketch) to be a dependent copy, or an independent copy. If Pro/E uses a dependent copy of that initial sketch, then it doesn't carry dimensions with it, just the outline. But if Pro/E uses an independent copy of the sketch, then there isn't a link. There is a solution to allow linking to the same section as needed at different points, but that wasn't ready for the current release.

- Three Ages of the Pro/E GUI

The confidence PTC people show about the Wildfire GUI has a solid foundation, a User Model. You don't usually hear of User Models when you buy a product, but any user interface (not just software) depends on a User Model, even if it's just in the mind of the designer. And PTC put man-years of work into their User Model, describing generally how people react with a CAD system to create and edit 3D geometry (a new area, not well defined). Someday nobody will buy any product that uses software (like, a cellphone) without asking to see the User Model, but that is a ways off.

Seems there are 3 ages in the history of the Pro/E GUI:

The Golden Age: from Rev. 1 to Rev. 19, the GUI was driven simply by the underlying code, the programmers wrote the GUI. Every time you hit a Done in the menu manager, it usually matched exiting a subroutine in the code itself. That was an amazing time, that for so long a period users and programmers could be in such harmony, and users could happily trace the structure of the code in their menu choices.

The Lipstick Age: this name comes from Jim Heppelman, and he used "lipstick" to describe the GUI development starting with R20. The whole emphasis there was to "slap" a Windows-like interface on top of the old menu manager, even though nobody ever suggested Microsoft knew anything about 3D CAD. The result wasn't a positive benefit, it was just cosmetic, just lipstick, Jim said.

Both of these first two GUI ages at PTC, the Golden age and the Lipstick age, did not have a User Model. But the sheer problems of trying to develop a real graphical user interface for Pro/E forced the development of a User Model, and that resulted in:

The Wildfire Age (no other name needed). And that's where we are now. This may be more centered on the graphical display itself than any other user interface around. And it has that User Model to guide it, which means it probably will look strange to users initially, just because users anywhere have rarely had an interface designed really for them (rather than an interface shared with the programmers, for example, or an interface that is nothing more than Windows compliant, for another example). A much bigger change than the cosmetic changes from R20 on.

Probably everyone involved with Pro/E has to be nostalgic sometimes for that Golden Age. It was so simple, how programmers would write code and then users would follow that code along in the user interface. Perhaps in the history of software applications there has never been such a fine match between programmers and users, sharing the user interface, as on R1 thru R19. But although Windows might have had some share in ending that Golden Age, with the R20+ changes, the real change probably is the development of a User Model to describe working with 3D geometry.

Something new there, and the results have to be different.