Re: HELP! PC connect Unix informix database by Openlink ODBC

The comment at the very end is the key reason for this responce. djames@statsci.com (Daniel James) wrote: :Nils Myklebust wrote: : :> Basically Windows 95 runs on anything from a 33 MHz 486 and up as long :> as it has at least 16 MB memory. With a price on the order of US$100 :> for all new memory it is hardly worth the trouble not to upgrade. The :> reason is that Windows 3.1/3.11 is so old and most software is no :> longer maintained for it. Windows 95 and NT also comes with buildt in :> TCP/IP that's reasonably easy to install and is also relatively :> stable. That could hardly be said of most TCP/IP software on Windows :> 3.1/3.11. : :Okay, I agree that the Win95 IP Stack is much better than the old :Winsock, however I wouldn't really want to run Win95 on anything other :than a Pentium or High-End 486 (like a DX4/100 or so). I've used Win95 :on a 486/DX266 and it was barely usable. Something streamlined and well :written like Linux or FreeBSD is usable, but Win95 is just too heavy. Sorry, this is plain wrong. It depends on what you try to run on the PC. Not everybody need to run the full MS Office suite. We have many users of 486/33 PCs running Windows 95 quite happily. The performance, with enough memory, is on the order of what it was with Windows 3.1 :> Better yet, if you can afford it is to run Windows NT. One key :> difference is that you can continue working with other tasks if some :> operations in one program takes a long time. : :My understanding is that Windows NT is what Win95 should be--stable :(although it requires a bigger footprint than Win95). I am not all that sure of the stability of Windows NT. I have had as many crashes of my PC runing Windows NT as I had with Windows 3.1. Granted all such crashing was related to running beta or even alpha software, but NT wasn't supposed to crash even if the application does. For general usage by end users it is stable though. :> Also if you want to run MS Access which is a good tool for ad hoc :> querying and reasonable for reportwriting you need to upgrade your PC :> enough that Windows 95 can be installed on it anyway. This is true for :> most other reporting and similar tools as well. To run even Windows :> 3.1/3.11 with only less than 16 MB has a tendency to result in very :> bad performance. : :Access is a nice tool for quick and dirty reporting, possibly even :not-so-dirty reporting, although I don't see a *requirement* to upgrade :to Office 95/97 for that capability. I've used both the 16-bit and :32-bit versions and they are both just as capabile a tool for this :purpose. Reread the above. I said you should upgrade to 16 MB to run MS Access (even the old 1.1 or 2.0 versions). If you do have 16 MB memory you can install Windows 95 and get about the same performance as using Windows 3.1. :> What kind of PC network you use (Lan Manager or other) doesn't matter :> for ODBC connectivity. That only uses the raw TCP/IP protocol. : :The OpenLink drivers also support IPX (if you really need to go that :route). Although since you're connecting to a Unix box, you'll probably :be using TCP/IP. To gain any advantage from using IPX it would mean your database was on a Novell server - a dying technology in general and not exactly the best platform for a database. But you are right of course. :> If you don't upgrade your OS you will loose a lot of time and effort :> trying to make Windows 3.1/3.11 work. It's not worth using the :> currently most pressious resource - human brainpower - on such an :> undertaking. That will cost you more than the upgrade. : :In my somewhat jaded opinion, migrating to Win95 won't eliminate a lot :of time and effort trying to get things to work, you'll just have :different tasks to undertake. I will admit that configuring the IP :stack is much easier under Win95, but once you have Winsock or whatever :stack you're using on Win3.x configured, the differences in making the :ODBC connection happy are negligable. This is where you, and seemingly many others, are so wrong. 1. The difference in configuring TCP/IP is significant and as you say much easier on Windows 95. 2. The stack available from Microsoft for Windows 3.11 (there was never any usable one for 3.1) had a number of bugs even in it's last incarnation that we kept hitting. 3. You can buy TCP/IP stacks from others that works more or less well. I never found any that where reasonably easy to install and not prohibitively expensive. We tested and used many and there where all kinds of problems. These all disappeared with Windows 95 and NT and TCP/IP network management became simpler. 4. The key issue: Although you will still have things to do using Windows 95/NT to get things to work, you will have the key benefit of beeing able to upgrade your software if there is a problem. There is little development of new software or new versions of existing software going on for Windows 3.1/3.11. Someone may counter this with several examples (even Java for Windows 3.1 and new ODBC drivers are still coming out :-), but the it's still the case that where new things are beeing developed is on Windows 95 and NT. If you have a problem with a software package you are using (any one on any OS) the first thing you ought to do is to check if a new version is out. If it is, you try it or otherwise check it, to see if it solves your problems (without introducing new ones). You don't allways use the latest and "greatest" of all software, but you do check new versions *when you have problems*. If not your losses in time and effort may cost you multiple times the cost of any upgrade. This by the way is as applicabel to Informix software on any platform as it is to general MS Windows software. (Just remember that according to the above it's still OK to use OnLine 5.x in many cases.) It seems to me that there are a lot of people that do not understand these issues. They keep working with old software, trying to solve all kinds of problems. They forget to calculate the costs. When you do calculate the costs you may first think of the hourly costs of developers. It may by itself warant the costs of upgrades. These costs are however commonly miniscule as compared to the advantages gained by using the time to develop new systems or functionality. Aren't people making these considerations? Nils.Myklebust@idg.no NM Data AS, P.O.Box 9090 Gronland, N-0133 Oslo, Norway My opinions are those of my company The Informix FAQ is at http://www.iiug.org