Re: Strw Poll : I-4GL for NT

Mike Segel wrote: > > David C. Harrison wrote: > > > > Mike Segel wrote: > > > > [snip] > > > > > I realize that this gets slightly off topic, but my point is > > > chosing the right tool for the right job. If you look at the > > > strengths of 4GL, and the few strengths of NT, you will see that > > > it is trying to fit a square peg in a round hole. > > > > I don't disagree with you on the subject of choosing the right tool for > > the job - but in at least one case I would argue that 4GL for NT *is* > > the right tool for the job. That job being to get an existing 4GL > > application to NT quickly. > > David, I have to concur with Mike, NT is simply the wrong environment for 4GL. 4GL is wonderful on UNIX, but trying to apply a UNIX-based tool, or server-based tool to NT is an abberration of NT, and 4GL. You are mixing cats and dogs. It is the wrong combination. NT in and of itself is terribly deficient even as a two-tier server. There is so much missing from the core OS. It is a miserable attempt to provide server services with less than most UNIX people expect. NT is a mess. Six months ago it was targeted to wipe out Novell. Now it's being targeted to wipe out UNIX, which is not going to happen. The only people who like NT are people who've never really used it. NT will also become more expensive in the next release. The propaganda from Redmond describing the initial price of the next NT was so badly received that MS decided to come out with a "lite" version of the next NT, for people who won't be able to afford the "enterprise" version. Excuse me, but why not try to finish one version before coming out with the next one. What a piece of shit NT is. And now the customer has to endure another "lite" version? When does this end? UNIX is a finished, mature OS. It comes with everything you need for a real server, except the data base. NT comes with high per-user licensing, enormous resource requirements, proprietary and sole-source vendoring, and an unstable operating environment. It is missing so much the only people who recommend it are amateurs. There are so many amateurs in management, that MS is selling the hell out of NT based on ignorance. > Sigh, > Well, if you want to take this into a philosphical discussion, we > can take it off line. (Mikey [at] Segel [Dot] Com) or > IM_Gumby [at] hotmail.com. > > But the issue isn't taking an application to NT. > The Informix database could reside anywhere. So if one > choses to use NT as the Sever, that's his mistake. > (Gee I wonder who supports Linux ;-) > > But to try and take the app to 4GL on NT, it wont work well, and > even if 4GL did exist, you would still have enough problems > that it would be easier to use a difference Client Server > paradigm. If you insist on NT, then use Delphi, VB or some > other client server tool. > Agreed. Even at that, NT is a miserable server to manage, yes, EVEN WITH INFORMIX. Informix runs pretty well on NT, if you spend a fortune on the server. It's amazing, but NT still runs slow on really fast servers. And NT has an interrupt-based model for processing. It's compared to UNIX all the time, but it's really more like VMS. > The reason I suggest this, not to bash NT, is that you have a > shift in the model. NT does not handle telnetd well, nor does it > handle other server issues well. I haven't seen an NT with 100 ttys > connected to it. Have you? Web server? OK. DB server? OK? App server? > forget it. Simpler and cheaper solutions exist. > Exactly. NT does not scale at all really. There is no real scaling model to apply to NT, as in UNIX. Wolfpack is a miserable attempt to try to upsize the processing, but it is not really a scaling concept, it's a clustering concept. UNIX can be clustered already, and XPS will work with probably just about any UNIX. You have a choice among several vendors to go with as well. Only one vendor for NT. XPS is targeted at big systems, however, there's no reason why we won't see XPS on smaller systems sooner than later. > Again, let us look at what the original poster wanted to do. > Port an application to a new platform. 4GL is not a good tool, > for other than a reporting tool, since it does not fit the > client/server model well. 4GL is mainly a server side tool. > (Yes it does have remote connections, but it is not the point.) > Well, in its day it wasn't bad for an app builder, and it still offers many back-office applications a great environment to getting data base work done. > 4GL is a non GUI language. Else why did Informix create New Era? > > Think about it. If you port 4GL to NT, what will you have? > An application which may work, but will not work well enough to > keep the users happy. You also have techincal issues which have > to be addressed. > Right. NT was originally pointed and targeted to a true client/server model. It was so hard in its definition, most people didn't get it. They immediately tried to turn it into a UNIX, because that's what they really wanted, a cheap UNIX on x86, with a great GUI environment. Guess what's come out now to fulfill that dream. Most sites I've been to try to turn NT into a UNIX. Softway Systems sells a UNIX-toolset for NT, as well as MKS. These tools are necessary to fill the gap for the miserable piece of crap NT is. Mike continues, talking about NT and 4GL: > Again, round peg in a square hole. (It may fit but not well). > > IMHO as a consulting software engineer, I would still maintain > that you would be better off doing the following: > > 1) Rewrite critical business logic in C > 2) Write a separate but equal application using whatever tool > you like in the PC world. As long as it can support C > > This will have the ultimate effect of reducing your maintennace > cost in the long run. Which is what is the major cost in > software. > > This allows you to place your core business logic on either the > client, or the server, or even the database as a C-SPL (C used > as the stored proceedure language.) > > As an added bonus, it allows you to rethink and revisit your > initial design, and utlimately improve upon it. > Informix talked at the IWUC about using server-side Java as the next wave in using server-side apps, and stored procedures. We're going to have to really think long-term on solutions, and Mike is right when you think long term. Informix 4GL may make a "come-back" if a low-cost UNIX were to take advantage of it. :-) > > Your whole argument is based on technology strengths and weaknesses. > > When was the last time the market chose a product on technological > > merit? Erm, I guess that would have been Windows, right? > > And your argument David, was technology based as well. It was/is based on shoe-horning 4GL onto NT, which offers a great short-term solution of a "working app", but long-term high-overhead. It's already higher- than-normal overhead for NT, putting apps that really don't belong there only complicates the matter. Maybe a strictly server-side-only 4GL app might be right,