Re: What's better: one lage .4gi or lots of little one's ?

I can verify what M. Dillard states regarding load times. We had a 700K Fourgen program that took, on a good day, 10 seconds to load. Once we achieved over 15 users on the system (HP9000/800 G30), it take as long as 2 minutes+ to load. We urged our vendor to compile it, and it now takes less than 1 second to load. Plus, our memory usage as dropped considerably. Users actually called us to thank us for the improvement to the system. That is how dramatic it was. In regards to your own question, it depends on how you want to support it. If you are using the r4gl to maintain it, you are going to have to keep a database around in order to 'compile' it. We currently are developing some .4gi programs because we do not have a full compiler (yet). I recently developed a function to handle the function of selecting print destinations. It made sense to break this function out into a seperate .4gl file because it could be used by mulitple programs. So multiple usability is one point to consider. Another question to ask yourselve is if you break it out into sub-programs, can each sub-program be tested or developed by itself?? If it can, then mulitple resources can be brought to bear, and testing procedures can be set up to test the modules instead of having to completly retest the full program when changes are made. I do not like too many modules because you start to impose constraints on the system. Your globals become large unless you fully think out what data each module is using and compartmentalize them. John Callaway Bath Iron Works My thoughs are mine alone, use at your own risk. ---------------------------------------------------------------------------- In article <490gc3$o8v@cssun.mathcs.emory.edu>, perryd@fourgen.com (Perry Dillard) wrote: >>>>>>>>>>> Perry's response to Hans Karl Winter <<<<<<<<<< > >Only use p-code for dev. environment because it has a good >debugger. For production env. use the compiled version of 4GL. >Once .4gi files get past a certain size, approx. 1meg, it starts >taking a really long time to load. Even on an HP-9000 machine. >For example, we have the following .4gi and it takes approximately >20 seconds to load. It is closer to 2meg, but the point where >the loading really starts to slow down I believe we found to >be around the 1meg mark. This is generally ok in the dev. >environment, but user usually want to see a better load time >than 20 seconds. > >-rw-rw-r-- 1 clintonc dev 1715771 Nov 22 10:01 i_ordent.4gi > >> From: hans@i2.com.br (Hans Karl Winter) >> Subject: What's better: one lage .4gi or lots of little one's ? >> Date: Wed, 22 Nov 1995 15:02:44 GMT >> To: informix-list@rmy.emory.edu >> >> Hi, >> >> I'm working in a big PCP program using: a Risk based HP , informix >> OnLine and Informix 4GL. I would like to know others experience in >> having large 4gi or lots of small 4gi. What is the real performance >> win ? What are the advantages of each technique ? >> >> >> Thanx in advance, >> >> Hans Karl Winter >> System Analyst >> EBC Sistemas >> --- Brazil --- >> >> >> >> > >-perryd (Guru) > >==================================================================== >= Perry Dillard - Chief Architect Direct Dial: (206)517-6430 = >= FourGen Software, Inc. FAX: (206)522-0054 = >= FourGen Building Tech. Support: (800)444-3398 = >= 115 NE 100th Street Sales: (800)333-4436 = >= Seattle, W.A. 98125-8098 = >= = >= e-mail: perryd@fourgen.com Web Site http://www.fourgen.com = >====================================================================