Are we crazy

Scott writes. >We have just found out that we need to develop a package very quickly. From >past experience, we believe that it would take about 8-10 weeks to do the >job in RDS/4GL (using our own 4GL code generators!). The powers that be >want it done in about 4 weeks and with Windoze front ends! (ie telnet session >isn't good enough!). Much of the tricky processing etc can be handled with >SPL and triggers but our experience in developing Windows Client Server stuff >is fairly limited (have done masses of 4GL stuff though). We need to work >out the best way of doing it. There will be plenty of "Non-trivial" screens >to write. The following is a list of some of the possibilities. We would >appreciate any thoughts on these options: > >1) NewEra >2) ODBC - Visual C++ } > Visual Basic } - They already own dev tools for these 3. >3) Powerbuilder } >4) Magic >5) Run away to a South Pacfic Island. >6) Other..................... > >The DB will have some tables with upwards of 1 million rows. There will >typically be up to about 20 simultaneous users (mostly simple quick queries >though some big reports). They have a 90Mhz Pentium running SCO Unix >and SE 5.02 available to run the show. > >Are we crazy?? > IMHO - Your apprehension proves that you are not crazy. The powers that be are setting you up for failure. They apparently want to prove that OO methods are not any good. If you think that this project will take 8-10 weeks using methods you are familiar with. Then it will take atleast twice that, to develope your first real project using OO. And that is assuming you have already selected, been trained in, and have played around with an OO tool. Your first OO project sets up the objects that will be reused in the NEXT project. The second project uses stuff from the first and adds more to your repository. As you move along, your projects get done faster and faster. But you already know this; you say you have home grown generators, and probably other stuff, that allows you to give 8-10 week turn-around. Two years ago (or even one year ago) could you have done the project in 10 weeks? Unless you want to milk your current employer for the training and experience that you will take with you to your next job, my advise is this: "Don't get involve with OO methodology, unless you have the full backing of management. Management must understand that not only is there a long learning curve, but the first few projects will take longer than when using traditional methods." : : Hmmm... Option 6... I put this mail away for a couple of hours and I've been thinking[look out!] I don't like saying "It can't be done".... Microsoft Access is very easy to learn and is relativly cheap (is this where your Visual basic is?). This will build your quick-and-dirty Goo-ey windoze application. You're suppose to be able to access other DBMSs too (hence the name I guess). This might give you the training and experience you may need in the future AND statisfy the 'powers that be'. This will give you a working proto-type that could be used to model a real OO project with something like NewEra, or maybe will do just fine as is. My two cents. -- uunet!kmart!jregep [The customer doesn't know what is needed... and neither does anyone else.]