Vault-OS : Update 18/11/2011

It happened like this :

1. My handler code for form POSTs was getting ridiculously complicated. I started to think, with about 70% of forms still barely implemented, I needed an easier way to do server side pages of web app without turning to some massive unfriendly OS specific code that would turn my tiny program into a 4 gigabyte install.

2. I decided to use the Boost::Asio network library to do automatic dispatching of GET and POST request through a templated tag system.

3. I tried for a week or so to get Boost::Asio to compile with Watcom 1.9. Close but no cigar.

4. I tried compiling under MinGW in Code:Blocks. Ten seconds later, it was like 7 warnings and one error. Fixed the error and compiled it straightaway.

5. Decided to ditch Open Watcom, bit of a problem doing the DOS 16-32 bit compile now. Will probably port over to DJGPP which is GCC compliant code 100% for MinGW. Meanwhile, working on the Win-32 version anyway. OpenWatcom was great but not worth some of the headaches it creates by being so incompatible even with ANSI standard C code. I was having to write more and more boilerplate code which is not Vault-OS, just basic support routines for HTTP web server app. I am aware that others have already done a lot of this work and I don't want to reinvent the wheel for everything.

6. Gaped in horror as I knocked out pages of my C code with a couple lines of C++ overloads. Same pages as before running on a much tinier fraction of code that is very readable and maintainable.

7. Now that I was using C++ again, found that the standard SQLite wrapper was way better than my C code in hiding the complexity, allowing me to do things like multitable form editing using foreign keys.

So two weeks detour, seem to be back on track. Using MinGW Code:Blocks package which is supposedly easily compiled under Linux as well. I would bet that this code will compile pretty easily to DOS using SoftSystems packet driver library for TCP-IP. Primary goal is Win-32 first, then everything else.