Branching out from the 3000’s home page

David Greer links the HP 3000 to new technology from a home page of history. The president of Robelle Consulting is celebrating his 20th year at the British Columbia-based supplier of utilities and programming tools, and counts his HP 3000 background well into the 1970s. Greer joined Robelle’s founder Bob Green in 1979 after the two had worked together on a consulting assignment. Greer began leading Robelle’s corporate direction several years ago once Green moved to a development site on a tiny Caribbean island. The products from these legendary programmers that made the company, Greer’s Suprtool and Green’s Qedit, remain at the head of Robelle’s expanding lineup after close to 20 years in the field.

If Robelle could only point to its two flagships, its place in the 3000 community would be assured, with more than 6,000 HP servers around the world running its products. But Greer has helped the company branch out. Robelle was the first HP 3000 vendor to have a home page on what was called the World Wide Web in 1994, a result of Greer’s explorations of the Internet; its What’s Up Documentation newsletter and SMUG glossary and guidebook made early appearances on the Web, part of a Robelle tradition of publishing. The company has a Windows presence in the programming editor market with Qedit for Windows, which counts Greer as its chief architect. Robelle still supports the products it created in the 1980s to help programmers in a next-bench fashion: Prose for document printing, Xpress for e-mail; and Howmessy for database maintenance.

More recently, the company has been looking in new directions, over the last few years expanding its knowledge and products to the HP 9000 and Windows markets. Its early experience with the Web and Windows NT will be taking Robelle toward Java solutions and perhaps XML, according to a new corporate press kit. We’ve begun to see Robelle as a bridge company for HP 3000 customers, with roots deep in MPE but branches reaching toward other platforms and the latest technologies. It has maintained its grounding in the 3000 while probing the breaking edge of automated development. We asked Greer how a company of only 20 people can do so much, dancing into new technology while staying close to the MPE partners that brought it onto the floor.

Robelle has branched out into platforms outside of the HP 3000. Why, and what do you do to keep as sharp about MPE as a company that’s completely focused on the HP 3000?

We have a lot of people with a lot of experience in MPE, and you have a lot of customers who are paying you support dollars to do a good job for them. That gives you lots of capability to look after them and lots of desire to stay focused.
But 3000s don’t live in isolated places, cut off from the rest of the world. We have a lot of cross-platform training, which actually makes you better. Our people with MPE experience now have a lot of Unix experience, network experience. Most of that stuff comes full circle because of the cross-platform stuff. I think you’re better serving your installed base by having a broader base to draw from. To be totally focused on the HP 3000 means you have to learning these other [technologies] anyway. We can learn it really well because we’re on the sharp end of the stick, supporting customers who are on those platforms, and some who are running both.

You’ve been in the Unix marketplace as well as 3000 shops for several years. Does it remain more complex to support your customers in non-3000 environments?

It’s partly the way Unix was designed. You have a lot of human-readable ASCII configuration files, so they’re pretty easy to change. On the other hand, they’re pretty easy to read, and figure out what’s going on. Our experience is so broad now that it’s hard to know what’s more complex. In figuring out stuff for Qedit for Windows I wouldn’t say MPE was necessarily easier. In some aspects it was more difficult, because it didn’t use standard APIs.

It still seems in talking to people that running large, high-end Unix systems with large numbers of users and big databases is still much more difficult than doing it similarly on a 3000.

Do your 3000 customers tend to rely on you to help them understand technologies on other platforms?

There’s usually a whole team [at a customer’s site] to drive new platform choices. I feel like 3000 people are almost not encouraged to come over; they’re told, “We’re the Unix experts, and we’ll handle this.” The 3000 people aren’t asking us for this advice because they’re not having to do it.

How do you keep up a broad base of products, including several older ones, with a staff of your size?

First and foremost is good software engineering and a lot of automated processes. Every weekend a single jobstream launches on our 3000 and by the time it’s finished on Monday morning, we’ve rebuilt every line of every product and run every regression test suite for every product.

We use a markup language to make sure there’s no errors in the documentation markup. We do that all for three platforms. The 3000 is still the master controller, but it triggers off events that happen on our Unix and NT systems. That jobstream reschedules itself for the next Friday.

That has a big impact on our ability to keep delivering stuff. A lot of our stuff sits on top of libraries, and if you change anything in the libraries, our processes catch any problems like unintended side effects. When we change Speedemon to support jumbo datasets, its not that HowMessy will absolutely be ready to go, but it probably will be. The underlying core technology is ready.

We use a single-source base for the Unix and MPE offerings. Qedit for the most part doesn’t know what platform it’s on; it calls a whole virtual layer. We went to object-oriented stuff before it had really caught on. It’s a philosophy.

We de-emphasized some things, and our bonus products we don’t enhance very much. We decided DBAudit fit better with a database vendor, so we negotiated a deal where Bradmark now owns that product. Xpress, our mail package, we wrote in 1984 because Bob and I worked at home and found the telephone a huge intrusion. When we saw the whole Internet thing taking off, we de-emphasized it and lowered the maintenance price. I’m still using Xpress, but before the end of the year of the year I will have switched.

Ever been tempted to do a product that didn’t have anything to do with the 3000?

Not so far, but I wouldn’t rule it out categorically. I’m spending most of my time doing market research now, getting closer to our customers and looking for new business line ideas.

You’re using Windows Scripting Host to release one version per month of Qedit for Windows. What promise does this tool hold for the HP 3000 developer?

I see its promise in the process automation area. You see messages on the Internet now that ask “How do I get my data out of the 3000 and into an Excel spreadsheet?” WSH won’t answer all the back-end stuff; you have to get it into a format that Excel will be happy with. You can do that with our product line, or with DataExpress and others. For those that want to grow their own, you finally have a scripting tool that will let you automate all that interface. Before, it was hit and miss, with DOS batch files or program it in Visual Basic, using VB to do little automated downloads into Excel. WSH is a nice little scripting language that lets you program any OLE automation object, and do it cheaply. It’s zero cost, bundled into Windows 98 or you can download it off the Microsoft Web site. Think of them like jobstreams, except they’re object-oriented and you’re going to have to learn something VBscript-like or Jscript-like.

We also do a lot of teaming and design of projects through Web pages. We have several huge internal Web sites with thousands of documents. We let team members know that updated scripts are on a Web page, so we share the same script, and use the same build process for getting the source code onto our PCs. You don’t tend to make too many mistakes, because you’re using a script to do a lot of the work, and the stuff that’s manual is documented, and it’s not too many steps.

Robelle was an early adopter of the Web, perhaps one of the first in the 3000 community. Why did you climb on board so early? Did you know even in 1994 that it would change the way we communicate?

We went live in June 1994. I’d spent six months that year researching the Internet open systems space. I’d been running into things called Gopher and Archie, and you could never find what you needed. We had a 9600-baud modem that was our connection to the ’net, and I’m trying to learn more, and I stumbled onto the Web stuff. At that time we were the 80,000th officially registered Web site. I could see after struggling with Archie, just for finding things and linking things together, this was pretty neat stuff. A lot of my browsing was done with the [character-mode] Lynx browser, and that bias has carried through. We tend to be an any-browser kind of company, so our Web site is equally accessible through Lynx.

I didn’t envision all the change from the Web. The change we’ve seen so far is minor compared to the change we’ll see in the next five to 10 years. For businesses in general, it’s like the Industrial Revolution compressed into 20 years. You have an entire culture moving from a mechanization age to an information age, and the Web is going to be one of the big enablers of that to happen. How a particular piece of technology, like the 3000, is going to fit in that, I don’t think we know. The 3000 has to interoperate in a Web world. People have been talking about Web-enabling. That’s the minor stuff. I don’t have a clear vision of where it’s going to go, but I wouldn’t be surprised if the Web is so standardized that it brings the promise of things like computers talking to computers a lot more.

In what kind of scenario would you recommend that a customer use an HP 3000 as their Web server for Internet service? For intranet services?

Whatever you’re going to go with as your application development environment will help decide what your back end or intermediate solution could be. It’s real important that Posix is [on the 3000], and so much of the open systems stuff is successfully ported to the 3000. A lot of the new stuff is still coming from the open systems space, so it has a good chance of running on the 3000: Perl, Apache, Java. And CSY [the 3000 division] is leading in a lot of these things. I don’t think that HP-UX is bundling Java, for example. CSY went for it, and said they will.

We use a whole variety of Web servers: A Windows NT server for in-office staff. Our Unix machine is our public Web server. Our 3000 runs as a Web server, because it’s the fastest way for us to publish certain internal information that’s in databases. That interfaces to our internal Service Request system.

What good will Java be to the HP 3000 customer base over the next year, now that HP is shipping a supported version of it with the operating system?

It is going to be more important to the platform, but I can’t answer about the next 12 months. It sends really strong signals that it’s bundled with the operating system and supported by HP. It’s just another piece of the puzzle. You’ve also got CSY’s Visage initiative for a recommended strategy for client-server, and that’s Java-based. Java is an enabling technology.

What does XML have to offer for the HP 3000 database administrator or application developer? Will you have a product that uses XML in some way?

I think it’s too early for XML. You know it’s going to be successful because Microsoft’s embraced it. I’m not certain I see it providing something to the database administrator they don’t already have. I don’t envision it being a mainstream technology outside of what Microsoft will do with it. When Microsoft takes on something and makes it a standard, it tends to affect everybody’s life whether you want it to or not. The problem is the browser deployment problem; Microsoft has put it in Explorer 5, but that’s not what most people have. In terms of public deployment of XML, that’s what’s going to hold it back. What I like about it is that unlike HTML, in XML you either have a valid document or you don’t. Sixty percent of Explorer is to handle invalid HTML code, and guess what to do with it.

You use the word proprietary when defining the HP 3000 operating system and database. Why is it important to describe them that way?

I’m doing a lot more work with outside consultants, people who have never heard of Robelle or the 3000. When you say proprietary HP 3000, it helps them put a little box and what kind of market we’re in. It’s a fact, like the proprietary AS/400. The world is moving to one of the most proprietary platforms that ever existed, Microsoft Windows NT.

I think it’s important to address the issue that people want solutions. Tools allow you to do huge amounts of customization exactly for your environment. We hope down the road there will be a Smith-Gardner Qedit for Windows scripting language part of our Web site, or Amisys or Summit or MANMAN. That is one of the ways we’re addressing the fact that customers say they want solutions, but we’re in the tools business.

Do you find it frustrating that your company’s legacy and legend has to be reintroduced to users brand-new to the HP 3000?

It’s frustrating, but you can’t rest on your laurels. We’re been around 20 years, but what does it mean? It probably means some of the programmers we’re trying to sell to were in kindergarten when we started. You have to remind them what it is you do, and move more toward solutions. People just don’t have the time, so you need to reduce their learning curve.

You’ve got a bigger staff than Adager, but a smaller one than Bradmark. What’s the optimal size for a company to stay creative and focused?

Go get your MBA and you still won’t know what the answer is. It’s not so hard to figure out what you should work on. The hard part is figuring out what you shouldn’t work on. We tend to be a fairly inclusive organization — we’ve had six people leave in 22 years. We have more of an international distributor network, and while they’re not employees, we think of them as part of the Robelle family. Adager has done more international business through Sun Valley. A lot of our distributors have been with us since 1980.

Every two weeks we have a company-wide staff meeting, and everybody has to produce a status report beforehand. It’s part of the glue that holds us together.

Robelle has published more almost anybody about the HP 3000. Why share so much of what you’ve learned for free with the community?

When we publish, we never give away the state secrets. To get the details right is bloody hard, and it makes a certain amount of sense to let Robelle do it. We’ve got the expertise to keep it up to date; CSY was very kind to give us early access to MPE/iX 6.5 and a high-end machine with a fair amount of disk space. But when you’re testing files above 4Gb, you can chew through 55Gb of disk space in an amazingly short amount of time.

You publish this stuff because you know how much hard work is really involved. What we’re sharing is the basic ideas, because we think it’s important to share them. It was a form of stealth marketing that’s been successful, a way to let people know that we have a certain amount of expertise, and we knew what we were talking about — and if you came and dealt with us, there’s a good chance we were going to look after you pretty well. We put a really high emphasis on quality, and trying not to promise what you can’t deliver.

It gets back to why our employees are creative. We like sharing this kind of thing, helping our customers understand.