Newsgroups: comp.databases Path: utzoo!telly!evan From: evan@telly.on.ca (Evan Leibovitch) Subject: Re: Opinions wanted on Empress Database Package and 4GL Organization: The Southern North York Spam Lover's Club Date: Tue, 9 Oct 90 01:26:56 GMT Message-ID: <27112761.EC@telly.on.ca> References: <1990Sep26.192938.1101@tropix.uucp> <96@thinc.UUCP> <270DDFD3.426A@telly.on.ca> <330@tslwat.UUCP> In article <330@tslwat.UUCP> louk@tslwat.UUCP (Lou Kates) writes: >In article <270DDFD3.426A@telly.on.ca> evan@telly.on.ca (Evan Leibovitch) writes: >>programs or distribute programs to branch offices, as EMPRESS isn't >>available in a run-time licence. Every site running EMPRESS applications >>needs a full-development licence. >>in Empress. >I don't know about Progress but we found that Empress' entire >development system cost less than the runtime of many other >vendors' systems Not at all the case. The Progress runtime for a 386 running Unix, unlimited users, is <$800 U.S. >and its often handy to have the development >system on the client's machine. This, I guess, is a matter of philosophy. Having development systems on each clients' system allow, even encourages, "quick fixes" which are not applied globally. The developer is saddled with the support nightmare of dealing with different clients, each of whom uses a version of the software on their system slightly different from the original. And I don't know about you, there are clients I have who I want nowhere near the ability to change the database schema/dictionary. >>Empress' M-Builder >>is very clumsy, uses non-standard mechanisms for describing terminal >>keys and characteristics (no termcap or terminfo), and is especially > >Since Empress runs on some non-UNIX systems, being able to set >these items up in a way which is independent of the way the UNIX >utilities work does have the advantage that it would allow you to >use identical terminal setup procedures on various versions of >UNIX and on non-UNIX computers. Sorry, but that's no excuse for Empress having such a miserable way to define terminals. Both Progress and Empress run on Unix, VMS, and MS-DOS (Progress, also on BTOS/CTOS). Progress uses a superset of termcap, while Empress uses a series of terminal database files in which each function key, escape sequence, screen function, and screen attribute is a separate record. Both use their respective terminal description methods across all platforms. But VMS, BTOS and MS-DOS systems rarely have to deal with a wide variety of terminals. It's mainly in the Unix environment where this is necessary. So why not borrow from the widely-used Unix techniques, rather than re-inventing wheels? (This probably has to do with the historical roots of each product. Empress was a VMS product ported to Unix. Progress was a Unix product ported to VMS. But the developers of Empress were quite familiar with Unix by the time they developed the M-Builder terminal interface.) I've had to write terminal definitions for both Empress and Progress. The Progress termcap approach is a winner before you start, because there are so many existing termcap entries floating around. One can run infocmp on any of the many terminfo definitions supplied by AT&T and come up with most of the requirements of a "protermcap" entry, Compared to this, the Empress method is indeed clumsy, and far slower to implement well. The approach to writing for termcap and terminfo is better documented, not only by AT&T but by third-party books such as the (highly-recommended) Nutshell book. The Empress docs, while adequate, are nowhere close. The terminal-configuring procedure, of course, wouldn't be a bother if the vendor supplied a complete list of supported terminals. But in the case of Empress, there was no definiton for something as basic as the Wyse 60, and I recall the AT386 console description being broken. -- Evan Leibovitch, Sound Software, located in beautiful Brampton, Ontario evan@telly.on.ca / uunet!attcan!telly!evan / moderator, rec.arts.erotica Of all the things I've lost, I miss my mind the most