>I decided to create my own installer... I do not want to use InstallShield, Inno and so on...
>If I create my own installer, I can manipulate the things during the process of installation time...
>
>3) So, lets supose that I just finished to format the machine and installed the SO (Win 98, for example)... Then If I try to run a VFP executable, it will not work because the DLLs files are missing. How can I solver this problem?

This sounds like a bad idea, Rodolfo. Like first, what language are you planning to build/run your installer with?

Obviously it won't be vfp, since you don't have the runtimes installed yet. You can actually still call vfp runtimes on a server or CD, but if you have any OLE/COM objects, for example, that won't work. VC++ would be my choice to build an installer, but you're starting a long & difficult project with any language, IMO, since you must assume you'll have to install the proper runtimes and any other files you need for your installer, just to get your installer started, before you even get to a setup process. How will you do this?

Then next, the registry modifications you'll need to implement are downright dangerous. What works on one OS/configuration often won't on another, and may be disastrous (we're talking about possible "hosed machines" here, and users don't like this :-)

I'm not saying you can't do it, but why on earth re-invent the wheel? Your time as a developer should be more valuable than creating something unnecessary - you can be creating vital, original, desktop/network/web apps instead, can't you? Trying to build something like an installer seems like a vast waste of resources, unless you can find someone to pay you for all the time you'll spend trying to get it right for all OS and configurations. That's a huge, long-term project. The risks and cost in lost app production time far outweigh the benefits, IMO.

The cost of a decent installer is not trivial, I admit, but compare how much time you'll need to spend building one (perhaps even years - and by the time you've got it right, it will likely be obsolete or need a major upgrade for a new OS :-) You'll spend the remainder of your life on maintenance, just trying to keep your installer current, supposing you do successfully build one. I'll wager your time is more valuable building apps, not building and maintaining installers. I know mine is, anyway.

My advice is: if you're unhappy with a particular installer, shop around, try WISE if IS is not to your liking, for example. Besides, installers like IS & WISE can be quite flexible (more or less depending on which version and how much you spend) - IF you learn how to use them, and learn where to get help when you're stuck (I just received some good advice right here, in fact).

Personally, I was a little nervous at first with vfp7 setups, after using the vfp built-in setup wizard for many years. But as it turned out, the learning curve wasn't bad at all, I was amazed how quickly I threw together my first complex vfp7 setup, including 8 merge modules and a fancy post-executable, all using Installshield Express (but not the one that comes with vfp7 - the one you purchase at a discount, which costs about $250). It's well worth the expense, IMO, after I floundered around a few days with the free version that ships with vfp. It may be okay for simple vfp apps, but you know the old saying, "you get what you pay for." So I also advise spending a little for a better installer than the freebie.

But good luck, anyway, if you go for building one. It might make for a good hobby, even if it's a bad investment of work time.