At 06:44 PM 11/25/00 -0400, Claudio Valderrama C. wrote:

>
>> If there is ambition to move FireBird forward, I think a conversion
>> to C++ is not only desirable, but essential. Or to put the
>> question another way, I think the effort to convert to C++
>> will return a major short term profit and a promise of long
>> term dividends ("dividends" -- an amusing but archaic financial
>> concept).
>
>Oh, an OOW (Object Oriented Wolf)
><g>

Guilty.

>Do you think that all the code needs to be rethinked or only retouched in
>the critical places?
>What's the state of C++ compilers in the N+1 platforms today? Can we rely on
>non-ultra-tricky code behaving the same on several compilers?
>

I think the order of battle would be this:

1. Change all engine .c files to .cpp files
2. Clean up the million warnings
3. Upgrade (i.e. replace) current exception handling
4. Gradually convert major subsystems to objects.

I suspect that the first three phases would take under a week.
A little experimentation will suffice.

I have great deal of experience with a, er, content store written in
C++ using gcc on Linux and MSVC++ 6 on NT. No sweat. I have it
on good authority that the jerk who initially hacked together,
ah, Firebird had a very similar mindset and coding style. So
I would say yes, you can expect non-ultra-tricky code to work just
about everywhere that gcc has already been. Having experienced
DEC's original C, I avoid ultra-tricky code with the zeal that I
use to avoid Dale.

Jim Starkey