Le Club Macintosh de Montrİal Publication: InterFace Issue: April 1992 Author: John Markle Series: NeXTdirections Title: Good news, bad news ------------------------------------ NeXTdirections Reflections on programming the NeXT computer by John Vincent Markle Good news, bad news The word "news" comes from the four directions: north, east, west and south. Business intelligence is news which is of interest to a specific corporation, news which can have an impact, positive or negative, on any one or more of that corporation's stakeholders: shareholders, managers, employees, suppliers, consumers, bankers, etc. Hence, by definition, business intelligence is either good news or bad news. This column hopes to provide developer-oriented business intelligence about NeXTstep, the object-oriented operating system from NeXT Computer, Inc. In my inaugaural column last month, I pontificated about the virtues of NeXTstep as a platform for custom application development. Consider that the good news. The bad news is that local experience has shown that programming isn't quite as easy as NeXT advertises, that the available documentation is inadequate for the neophyte developer, and that it is difficult and expensive to obtain assistance when one is stuck or confused. To program under NeXTstep, the average programmer must first adjust his thinking from a function-oriented paradigm to an object-oriented one; this is no mean feat, even for the professional. Secondly, one must learn the nature, features and methods of the many objects in at least the Application Kit and the common classes, and learn how to construct networks of objects, including new ones, using the Interface Builder. Thirdly, one must learn to program in at least Objective C and Display Postscript, and optionally learn the basics of compiling and linking within a Unix shell. Finally, for efficiency purposes, one should learn to use GDB, the GNU debugger from the Free Software Foundation, as well as the other debugging tools currently included within NeXTstep±AppInspector, MallocDebug, and ProcessMonitor. The local consensus is that the documentation provided to assist in this laborious journey could be considerably improved; at the very least, a simple road map should be supplied. Third party books do help to decipher the many manuals and documents about NeXTstep and its related programming languages, but NeXT itself should be taking more direct responsiblity to support what is currently its primary market niche, custom application development. If one wishes to receive programming support from NeXT, one must first attend the Developer Camp ($1,500US tuition, plus travel and accomodation expenses), and then sign up for the Technical Support Hotline ($2,500US per annum). The dealer network can't be expected to provide programming support without direct financial compensation, and so many store-front dealers lack the requisite technical expertise to provide such support. Another problem is that NeXT uses Internet for its inter-site communications, hence handicapping all programmers who are not in a university, the military, or a Fortune 500 company. Nevertheless, in spite of these serious drawbacks, NeXTstep is, in my opinion, the most exciting development platform around, the most challenging to a professional programmer, and the most in tune with the future. However, until the above-mentioned problems are rectified, the NeXT will remain a machine by and for an elite. Getting back to good news, success with RAD techniques (Rapid Application Development) has warranted a headline in the Macintosh world (MacWEEK, January 27, 1992). Popularized by James Martin & Co. of Reston, Virginia, RAD techniques are based on the premise that the best way to satisfy corporate user's application needs is to build small working prototypes rapidly, and then refine them. Small SWAT development teams (Skilled With Advanced Techniques) meet frequently with users and work quickly to build extensible systems with graphical user interfaces. Users get prototypes early, try them out and offer feedback; these prototypes are then refined and integrated in a second and third cycle. Hence, users' expectations are usually met when RAD techniques are employed, and the resulting systems can be easily extended to accomodate future needs. RAD techniques are a natural affair under NeXTstep, since interfaces and other system modules can be built and refined in record time; there is already plenty of testimony to this effect from the academic community. Corporations are picking up on this feature of the NeXT, and now business success stories are being told which involve NeXTstep; I will report on these stories regularly in future columns. Finally, here is some long-range business intelligence, a faint signal whose impact cannot yet be fully evaluated. On March 3rd, 1992, the New York Times reported that Sun Microsystems, Inc. had hired Boris Babayan, the 59-year-old supercomputer wizard of the old Russian military, and his team of roughly 50 computer scientists to open a laboratory in Moscow. (Sun, as you may recall, is now considered by Steve Jobs as "the mother of all competitors.") Boris' presence in the workstation poker game should raise the stakes considerably, since the Russians have long held the theoretical high ground of computer science; metaphorically speaking, Boris has a few cards up his sleeve as he sits down on Sun's side of the table. Those of you who are making odds on NeXT will eventually have to adjust your scorecards to accomodate the Babayan factor. Next month: To leapfrog, or not to leapfrog? [Interface, April 1992, Column #2]