HP 3000s drive some of the world's biggest databases. Our roundup of available reporting tools can make data work harder for your company.

Review by Cortlandt Wilson

It’s hard to imagine supporting an active user base without a fourth-generation language (4GL) reporting tool. For a good many HP 3000 users, 4GL means PowerHouse Quiz. It was the first 4GL many of us ever saw on a 3000. Unfortunately, Quiz has been showing its age for a long time. There are many other options with more user friendly interfaces — a product aspect that puts reporting capability into the laps of users. In my opinion, this is where a lot of that capability belongs.

In this review I examine every HP 3000-based 4GL reporting system with the exception of HP Transact and the 4GL Speedware. As of the deadline for this review, Information Builders didn’t respond to inquires about FOCUS. Quiz is listed first, as a familiar base line. The remaining products are listed in rough order of preference. At times I am comparing apples with oranges, so the order is necessarily approximate.

In this review I refer to system repositories (or simply “the repository”) instead of data dictionaries. (See sidebar). The reason is that a repository can store much more than just data definitions. A repository can store code segments or macros — giving Quiz USE files, for instance, a better home. Reports can be stored and centrally managed in a repository.

A repository is a place to store and manage reusable report elements. A powerful and fully stocked repository becomes a kind of CASE tool which automatically generates a lot of the report for you. End user reporting — the ability to turn the data and the responsibility of writing reports over to your users — is especially dependent on a good repository. The data dictionaries available for QUIZ, for instance, just hint at the capabilities of a mature system repository.

The downside of a repository, if you could call it that, is it they must be populated before they can be used. Populating or programming a repository is a bit harder than just coding a report, especially the first time you do it — just like designing a good database is harder than creating a new file on the fly. This is a matter of management economics, not total effort; pay me once now, or pay me later — every time you need it.

Some products don’t require a repository when reading IMAGE databases. Not requiring a repository for IMAGE database access was cited as a feature both by vendors who offered one and by those who didn’t. This may be good for quick access, but I believe it’s counterproductive in the long term. Many HP 3000 users won’t appreciate this — because we haven’t been exposed to the benefits of a really good repository.

Going beyond basic data formatting, repositories can manage SQL-style views which present selected data from several sources with the linkages transparently defined. User-defined data types, and even virtual calculated fields, can be defined correctly one time and then reused forever.

I want a repository that is a place to store reusable 4GL objects and the meta-data on how to automatically glue them together. A user-defined data type is an example of a reusable object. I want to define a MANMAN date conversion once, and assign it as the default display format for all my MANMAN dates. Among all the ideas put forth for improving programming efficiency and quality, re-use is one of the few that has proven itself as worth the effort. Experience has shown, however, that successful re-use requires some technical and management infrastructure to support it. If there is a ‘right’ way to write business reports, a high level language linked to a repository of metadata and reusable objects would be near the top of the list.

The first, and probably best-known reporting software for the HP 3000. Often considered the “default” report writer for the HP 3000. Sold separately or as part of the complete PowerHouse family of QUICK screen generator and QTP batch update modules. Runs on a number of platforms.

PowerHouse-related tools include the PowerHouse Web Server, PowerHouse Client, Axiant 4GL, and PowerHouse Power2000. The Web Server is a modified version of the QUICK module. Axiant 4GL is a client-server environment that can build basic client-server applications from Quiz, QTP, and QUICK source code. PowerHouse Client gives PowerHouse a more graphical, client-server front end, but is less powerful than Axiant 4GL. Power2000 is a Y2K evaluation and remediation utility.

Repository: Neither the older QTP nor the newer PDL data dictionaries support views or calculated fields.

User Interface: HP 3000 character mode. Blank slate high level language, usually written on an editor and executed in Quiz. There was a companion tool, rarely purchased, designed to help with report layout.

Input: All HP 3000 file types. Self describing files, ‘subfiles’ from Quiz or QTP. IMAGE with Omnidex. ALLBASE, Oracle, Sybase SQL Server, Microsoft SQL Server, Borland InterBase.

Output: Reports and PowerHouse self describing subfiles.

Analysis: Quiz is the utility that the competition has to beat, and in one way or another they do. Quiz’s system code is said to be difficult to maintain. The Year 2000 compatibility features added to the recently released version 8 were kept to a minimum because the designers were afraid of breaking something.

As the Web Server is based upon QUICK, it can’t Web-enable existing Quiz reports. PowerHouse Client and Axiant 4GL would be of limited value to Quiz-only customers. Axiant does share a common code base with the latest version of PowerHouse.

History: Originally designed as internal tool by a consulting group. A young consultant named Birket Foster received permission to sell Quiz as a product. In Foster’s telling, Quiz was the first 4GL that many had seen and it sold about as fast as he could demonstrate it. Later the QUICK and QTP modules were purchased and integrated into the PowerHouse family. The ad hoc history of the product family may explain why parts of it appear to have been designed after the fact.

Legend has it that a Cognos sales manager sold ASK the right to resell any number of copies of Quiz to MANMAN users and to get a portion of the support fees essentially forever. The sales manager made his quota and was promptly fired. ASK believed that its bundling of Quiz helped put the product on the map and got a lot of users under support. Cognos saw itself locked into a unfortunate contract with ASK, which forced it to give a lot of product away. Both were right.

Eventually Cognos significantly revamped the product with a new data dictionary. Cognos claimed that the contract covered the QDD dictionary and not the new PDL dictionary, therefore rendering the old contract moot. What ensued has been described as a battle of the titans and corporate egos. ASK resolved the issue by dropping Quiz support and adopting UDMS as its recommended report utility.

Platforms: HP 3000, HP 9000, VAX and Alpha OpenVMS, IBM’s AIX.

Pro: Well-known language. Runs on a number of platforms. Large support base, stable company with newer products doing well in the marketplace.

Con: The language has its quirks and glitches. Like all script-based tools, laying out the format of a report is tedious. Not end-user friendly. Weak repository. Inconsistent syntax across the tools in the PowerHouse family. By the criteria of repository in the Glossary Sidebar, the data dictionary for Quiz leaves a lot to be desired.

DataExpress takes the prize for reading the most diverse mix of file types, system repositories and data sources. M.B. Foster Associates (MBFA) is working to enable universal data across all the HP-supported operating systems — MPE/iX, HP UX, and NT. DataExpress is looking to become, if it isn’t already, the most flexible and “open” solution of the lot.

MBFA markets DataExpress and its ODBC solution as a single solution. ODBCLink, MBFA’s full-featured ODBC product, is only sold as an option of DataExpress. HP licenses a limited-feature version of ODBCLink and includes it as part of the MPE/iX fundamental operating system in version 5.5 Express 3 and beyond.

Repository: DataExpress can read metadata from PowerHouse’s data dictionary, HP’s Dictionary/3000, IMAGE root files, and its own repository. DataExpress’s own repository, some parts of which are sold as options, is one of the best on the HP 3000. SQL style views are supported.

User Interface: HP 3000 character mode. Prompted dialog. User-friendly features.

Input: Several types of self-describing files from Quiz (the above-mentioned subfiles), Suprtool, and HP format. Native support of Omnidex and Superdex through HP’s third party interface (TPI) standard.

DataExpress currently has had the ability to pull data from Oracle databases. A soon-to-be-released feature will add a more generic capability to read ODBC data sources from non-HP 3000 (HP UX and Windows NT) systems. MBFA has also announced an intention to support the OLE DB standard when such data sources become available.

Output: Reports, graphs, several types of self-describing files — including PowerHouse (Quiz) sub-files, and extracts in a number of Intel based PC and Macintosh formats.

Analysis: MBFA tries to provide a complete reporting solution as indicated by the DataExpress features that are in the works. These features include access to non-HP 3000 ODBC data sources, DataExpress UX on the HP 9000, and support of the emerging OLE DB standard for access to non-relational database data. The various ODBC vendors have given other platforms access to IMAGE data but not necessarily the other way around. The next step is the ability to read data directly from ODBC data sources from other platforms and easily join and process that data on the HP 3000.

Experience has shown that a solid ODBC connection will probably require some customization for each combination of ODBC data source and client. Short of a special arrangement, MBFA can only be expected to support data sources commonly found on HP UX or NT. Given that caveat, however, DataExpress will soon have the potential to access data from any ODBC data source on your network, combine it with MPE/iX data sources, and process it all on your HP 3000. It’s hard not to get a little excited about that.

History: Developed in the mid-1980s by David Dummer and sold under the name IMAX. Dummer, who now works for MBFA, also wrote the Rapid reporting system (which includes Transact) and sold it to HP. MBFA purchased IMAX in 1989 and renamed it DataExpress. Using the data access technology in DataExpress, MBFA later created their ODBC listener — the part of ODBC that runs on the HP 3000. In 1996 HP licensed the rights to distribute a limited version of MBFA’s ODBC solution as part of MPE/iX.

Platforms: HP 3000, HP 9000 (planned).

Pro: Broad feature set, interoperability, and multiplatform capability. Good user-friendly interface for a terminal based system. Powerful repository that sets the standard for others to follow.

Con: No client based interface. The included ODBC is only one of several on the HP 3000 market, and the capabilities of these products vary.

This is the first client-server report utility examined for this review. Like ISSI’s Safari and MBFA’s ODBCLink, Visimage is based on existing HP 3000 server technology. On the server, Visimage consists of a listener program and AskPlus, which processes the report requests. AskPlus can run reports directly from the HP 3000, as a stand-alone program. AskPlus is bundled with every copy of Visimage.

Multi-pass reports can be bundled into run packages called contexts. Recently, the capability to run Suprtool was added to contexts. The run instructions in a context can be translated into HP 3000 jobstreams.

A notable feature of Visimage is its pre-defined macros. The macros operate something like stored procedures in SQL. Macros can be written using Visimage’s calculations and functions, or they may call user-supplied programs with the ‘user exit’ facility.

User Interface: Client-server.

Repository: User-written macros are cataloged and can be called either as functions or as virtual data fields. Fields can be renamed and redefined. Other than this, capability is minimal. Can read its metadata directly from IMAGE root files or Allbase/SQL DBE files.

Input: IMAGE, KSAM and sequential files, Allbase, Oracle. Self-describing files in Suprtool link or HP SD formats.

Output: Reports, self-describing sequential and KSAM files, PC formats. Graphing and advanced data analysis must be performed by downloading to a third-party tool.

Analysis: Vital Soft recommends Visimage for HP 3000 reporting and Brio Enterprise (so far) for ad hoc reporting, OLAP data analysis, and data warehousing from multi-platform environments. This may partly be a matter of Vital Soft’s company focus, as Visimage does run on other platforms and reads Oracle. Visimage leads off with such a strong feature set that its weak repository features came as somewhat a surprise to me. No textual data descriptions; macros cannot be assigned to fields eliminating the possibility of transparent, user defined data types, no SQL style views.

Security is almost non-existent. Limiting access to particular files or fields can only be done by building a new dictionary. There are, however, some workarounds. Standard reports can be stored on a network server; the same data that would be directly obtained from an SQL-style view can be incorporated as pre-written extracts in the first pass of a “context.” Suprtool extracts can be incorporated into multi-pass runs in contexts, but I would prefer to see that technology transparently built right into the product.

In client-server mode, the user only interacts with the Visimage client on the PC. In stand-alone mode, AskPlus can be run as a host-only process. In this mode, the Visimage client can be thought of as a CASE tool for developing HP 3000-based AskPlus programs. This dual personality offers some of the best of both worlds.

The majority of Visimage HP 3000 users in the US are companies running Amisys HBOC medical management or Summit credit union management packages. These systems are entirely IMAGE based, and most users do not use the repository but rather take their metadata directly from the IMAGE root file. Vital Soft estimates there are more than 500 installed sites in the US, with many more in Europe. The European market is more focused on the HP UX version. An interesting side note is that Visimage worked with Robelle to develop the format for Suprtool Link self describing files.

History: R&D for Visimage is performed by the French HP 3000 vendor Ares. Vital Soft is the US sales and support representative for Visimage. Both companies are privately held, but there is some joint ownership. Ares was founded (originally under the name Cogelog) in 1977 by former HP employees. The new company purchased the rights to HP’s QUERY and significantly rewrote it. Ares released the host-based reporting tool AskPlus in 1985. Vital Soft was formed in 1987. Visimage was released in 1989 with a DOS-based client, and the Windows version made its debut in 1994.

Host Platforms: HP 3000, HP 9000, NT. Client Platforms: NT, Windows 98, 95, and 3.1.

Pro: Client-server and host-based modes of operation. Predefined macros. Available for both HP 3000 and HP 9000 platforms and Oracle databases.

Con: Disappointingly weak system repository, Suprtool-like technology not built-in. Cannot combine data from HP 3000 and HP 9000 into a single report.

Good reporting features, system repository, update capability, even simple screens. However, its implementation on the HP 3000 leaves something to be desired. The HP version is a port from the Digital VAX version.

User Interface: UDMS uses a VT 100 terminal protocol, causing the terminal or terminal emulator to switch modes when starting or leaving UDMS. A mix of prompted dialogs and blank slate programming languages depending on the module. User-friendly features in some modules.

Repository: Supports views, calculated fields, security, and report cataloging.

Input: All HP 3000 file types. ORACLE, INGRES, Sybase, Informix, Rdb, DEC-DBMS, RMS ISAM, Acucobol’s ISAM, and more.

Output: Reports, a number of PC formats, label formats.

Analysis: Compared to Quiz, UDMS has a number of powerful features, including a high-level 3GL language that supports looping. The system repository, once you figure it out, is one of the best on the HP 3000. The VT 100 protocol means the user interface has a somewhat unfamiliar feel and control functions. UDMS has a promising design, but its execution leaves a lot to be desired.

Platforms: HP 3000, VAX VMS, various UNIX, NT.

History: UDMS was ported to the HP 3000 with some reluctance at the behest of the old ASK for MANMAN sites. When ASK lost its sweetheart deal to resell Cognos’ Quiz, ASK went looking for another 4GL that it could resell along with HP 3000 and VAX versions of MANMAN. A number of VAX MANMAN users were already using UDM. ASK told ISSI that if UDMS was ported to MPE/iX, then ASK would market UDMS to all new MANMAN customers.

Pro: Reporting, extract, update, and inquiry capabilities. System repository roughly equal to DataExpress.

Con: Support for HP 3000 port, documentation. The documentation for the more powerful features of UDMS is rotten. It reminds me of the PowerHouse QTP and QUICK documentation prior to version 7. Several years ago I identified 16 syntax issues for a 3GL command that were either missing, ambiguous, or wrong. ISSI’s support manager agreed that 15 of the 16 issues were substantial. To my knowledge ISSI is still issuing the same flawed documentation for that command.

Safari is a true client-server system that uses an HP 3000 host listener based upon technology developed for UDMS.

User Interface: Client-server. Prompted dialog. User-friendly features.

Analysis: Safari, which was nearly two years late, is the company’s second attempt at a client-server companion product to UDMS. The MK Group recommends Safari to MANMAN customers as their low-level, end-user reporting option.

Platforms: HP 3000, VAX VMS, various UNIX, NT.

Pro: End-user friendly. Innovative features, especially in the system repository. Host server shares code with existing host-based UDMS reporting software. Available for a number of platforms.

Con: In the past ISSI treated its HP 3000 port of UDMS as a secondary product. Documentation was poor.

A low-cost tool with high speed, reporting, data extraction, and a batch update language.

User Interface: HP 3000 character mode. Prompted dialog with scripting language. Visual layout of report. Intended for programmers and power users.

Output: Creates KSAM self-describing subfiles so you can link anything to anything. Capability to add or update existing data.

Repository: Basic. Can read its metadata directly from IMAGE root files.

Analysis: About 200 copies sold. Now sold with single-tier pricing of $5,000. The batch update capability is a bonus. The vendor says that DataNOW! is often used for data conversion. The low cost and update capability makes the system attractive for single-purpose, limited-time use. Shawn Gordon’s review in the NewsWire’s October 1998 issue said, “DataNOW! isn’t the most glamorous looking HP 3000 application, but it is a very functional one.”

Platforms: HP 3000.

Pro: Price. Speed. Update capability.

Con: Weak repository. Support for only HP 3000 platform.

A powerful reporting language based upon the paradigm of a spreadsheet. Features include Postscript graphics, spreadsheet-like, near instantaneous recalculation, optional client-server front end with HP terminal emulator, speed, and efficiency.

User Interface: HP 3000 character mode or optional QCTerm client-server front end. QCTerm combines a PC-based WYSIWYG design environment and an HP terminal emulator. Prompted spreadsheet-like dialog. Visual layout of reports. Intended for programmers and power users.

Input: QueryCalc uses all the standard data access tricks and a few of its very own. The Web site describes a method of pre-indexing data which, to my knowledge, no other commercial system uses.

Output: A unique feature of QueryCalc is its support of the Postscript graphics language.

Repository: Minimal capability.

Analysis: QueryCalc is an odd duck among its competition, but one that I believe deserves more attention. The system is the brainchild of Wirt Atmar, a university professor with two Ph.D.s who turned his attention to developing high quality software for the HP 3000. QueryCalc shows that academics can come down out of the ivory towers.

QueryCalc is based on the idea of a spreadsheet, a concept which on closer inspection turns out to make a lot of sense for writing reports. Like a spreadsheet, you can change one item in a report and almost instantly see the results. In QueryCalc, you build a report by creating a heading sheet, detail line sheet and so forth, then tie them all together with a master report sheet. One programmer said he was uncomfortable with QueryCalc because he was unfamiliar with the spreadsheet paradigm. On the other hand, the same programmer said that his spreadsheet-literate users went to town with it.

Support of Postscript printers means that professional-quality forms and reports can be designed at the same time. The QCTerm front end provides a near WYSIWYG display of the final report.

The lack of features in the system repository is one of the few weaknesses of QueryCalc. Weakness of the system repository aside, QueryCalc gets my vote for the most powerful 4GL reporting language on the HP 3000.

AICS believes that 4GL languages should be mostly intuitive. QueryCalc is sold with a 10-minute rule — if you can’t figure out how to do what you want in 10 minutes, then call AICS Research for help. If that means you call several times an hour, then so be it. My experience over the years is that AICS answers questions quickly and accurately.

AICS Research doesn’t spend a lot of money on marketing. Those low marketing costs mean a very reasonable single-tier price and a run-only version for $500. QueryCalc combines a refreshing, unusual commitment to quality, data access efficiency, intelligent language design, graphics capability, and great support — all at a very reasonable price.

Host Platforms: HP 3000. Client Platforms: Windows 3.1 and above.

Pro: Powerful and flexible language, intuitive spreadsheet-like paradigm, built-in graphics capability, speed and efficiency, great support, QCTerm front end.

Con: Weak system repository, HP 3000 only. The Lotus-like macro language clone may be less familiar or cryptic to users of other spreadsheets.

The power and flexibility of ODBC-based reporting tools may be the right choice for some uses — especially where user-friendly ad hoc reporting, OLAP data analysis, data warehousing, and access to data from a number of sources is important. There remains, however, a place for high-volume, centrally controlled reporting software packages, so systems that allow for both client-server and host-centric modalities get extra points in this review.

DataExpress wins points for its broad feature set, user-friendly terminal-based interface, and multi-platform capability. Visimage and Safari feature user-friendly, client-server front ends. DataNOW! is inexpensive, fast, and even includes an update capability; it is handy for special one time uses such as data conversion. QueryCalc shines with its built in graphics capability, innovative spreadsheet-like interface, power and elegance. Behind the scenes, it may incorporate the best set of access performance tricks short of installing Omnidex or Superdex TPIs on your IMAGE databases.